Search public court records for Lafayette Parish (15th Judicial District Court) and Lafayette City Court in Lafayette, Louisiana, and explore aggregate statistics on how the courts process cases. Records originate from the Lafayette Parish Clerk of Court and the courts; this is an independent project and is not the Clerk of Court or any court.
This page also documents the project's methodology — where the data comes from, exactly how each number is calculated, what is missing, and where the limits are. If a figure on the site seems surprising, the explanation is here.
This is an independent civic-transparency project. It collects records that are already public — criminal case records from the 15th Judicial District Court (Lafayette Parish) and violation records from Lafayette City Court — and makes them searchable in one place, alongside aggregate statistics derived from them.
The goal is to let residents, journalists, researchers, and advocates see patterns that are hard to see one case at a time: how long cases take, how often charges are dismissed, how the pending caseload is aging, and how the system processes cases overall. It is not affiliated with any court, the Clerk of Court, the District Attorney, or any law-enforcement agency.
All records originate from official public sources. The data is read from the courts' own published case information and the documents filed in those cases — the same records any member of the public may request or view.
As the Lafayette Parish Clerk of Court itself notes, filings made with the court may become public record subject to the Louisiana Public Records Act. This project works only with that public record.
The Journey tab connects records across all five sources to show a person's path through the criminal justice system in a single timeline — from a sheriff arrest report, through a jail booking, to a court case filing, disposition, and any subsequent appeal. Records are linked by matching booking numbers, agency report numbers, court case numbers, and name plus date of birth. Some links are confirmed automatically with high confidence (a jail charge's docket number matching a court case number); others are based on name and date of birth matching and may be reviewed manually for accuracy. The timeline is a research and transparency tool — it assembles public records and does not indicate guilt, conviction, or any legal finding.
The Heatmap tab displays Lafayette City Court violation records on an interactive map. Each record that includes a street address is geocoded using the U.S. Census Bureau Geocoding Services API, and the resulting coordinates are plotted as a heat layer using Leaflet. The map supports filtering by year or a rolling time range, and clicking a location shows the individual records nearest that point. Not all records have a usable address — the data-coverage panel on the Heatmap tab shows, per year, how many records were geocoded versus how many are missing an address. Geocoding is an ongoing background process; coverage improves as new records are collected and previously ungeocoded records are retried.
The Jail Statistics tab in the search tool presents aggregate figures derived from the Lafayette Parish Detention Center inmate list: booking volume by month, demographic breakdowns (gender, race, age at booking), length-of-stay distribution for released inmates, top charges at booking, bail distribution per charge, bookings by arresting agency, and a repeat-booking rate. A live panel at the bottom shows the current in-custody population by charge, refreshed every 60 seconds. These statistics describe jail bookings, not court outcomes — a booking charge may differ from any charge later filed in court, and a booking does not indicate conviction or guilt.
For the authoritative version of any record, or to look something up directly at the source, use the courts' own sites:
The statistics are not assembled from headlines or summaries — they are backed by the underlying court documents. The project retains copies of the source documents it has collected so that figures can be traced back to their origin, and so the corpus exists as a durable public-interest archive of records that are otherwise scattered and hard to access in bulk.
These numbers grow continuously — collection of historical records and documents is ongoing — so they are shown live rather than as a fixed figure, and will be higher each time the underlying archive expands.
The project is willing to share source documents and underlying data with journalists, researchers, academics, and community advocates working in the public interest. If you are doing accountability, reporting, or research work and the records would help, get in touch at [email protected] and describe what you need. Requests are handled individually; the aim is to support legitimate public-interest use of records that are, after all, public.
The statistics are not a frozen quarterly report. They are computed directly from the current data, and recomputed at most once an hour — the aggregate queries behind them are expensive, so the results are cached briefly rather than recalculated for every visitor. Each statistics view carries two timestamps: a “data as of” stamp showing when the underlying records were last collected by the scrapers, and a “figures computed” stamp showing when those records were last turned into the numbers on screen. The two are deliberately separate, because a record collected five minutes ago may not yet be reflected in a figure calculated fifty minutes ago.
Because collection runs continuously, numbers move: new cases appear, pending cases resolve, and the document archive grows. A figure read today may differ slightly from the same figure read next week. That is by design — the dataset reflects the live state of the record, not a snapshot.
The definitions below are the exact rules behind the numbers. Where a national standard exists, the project follows it, so the figures can be compared against other jurisdictions that use the same measures. The two reference frameworks are the National Center for State Courts (NCSC) CourTools measures and the case-flow conventions used by court-data transparency projects such as Measures for Justice.
Cases disposed in a calendar year as a percentage of cases filed in that same year. Filings and dispositions are counted as two independent flows, each by its own date — a year's disposed cases may include cases that were filed in earlier years.
This measures how well the court keeps pace with incoming volume. It is not the fate of a single year's filings. The rate can exceed 100% in a year when the court resolves more cases than were filed.
A case counts as completely dismissed only when every charge on it was dismissed or nolle prossed. On a multi-charge case, a single charge with any other outcome — a guilty plea, a conviction, an alternative resolution — disqualifies the case. This is a case-level figure and deliberately differs from the charge disposition breakdown elsewhere on the statistics page, which counts every dismissed charge including counts dropped on cases that ended in a conviction on some other charge. In City Court, where each violation is its own record, the same rule applies to the violations sharing a case number.
Only cases where all charges carry a recorded disposition are eligible: a case with any charge still pending is not finished, so it cannot be called dismissed and is excluded from both the count and the denominator. The denominator is therefore fully adjudicated cases, not all cases filed. The percentage answers: of the finished cases from this cohort, what share ended in total dismissal. Year cohorts are keyed to the filing year, not a disposition year — a case is one unit and its charges may resolve across several calendar years, so there is no single disposition year to file it under. Rates resting on fewer than 5 adjudicated cases are suppressed.
A dismissal is not a finding of innocence, and this figure does not distinguish among the reasons a case ends: insufficient evidence, a witness who did not appear, a negotiated resolution, a diversion program completed, or a charging decision reversed all appear the same way in the record. The court record shows that the charges were dropped, never why. Recent filing years are thin and their rates volatile, because most cases from those years are still open and therefore excluded — the ones that have finished are the fastest-resolving, which is not a random sample. Cases dismissed before they were ever filed are invisible to court records entirely. Because the underlying numbers depend on how dispositions were typed into the court system, an unrecorded or misspelled disposition can move a case out of this count.
The average and median number of days from a case's filing date to the disposition date of its charges. Reported overall and broken out by case class (felony, misdemeanor, traffic), because those dockets resolve at very different speeds.
Same-day dispositions are excluded from the time-to-disposition averages. Only cases with the required dates present are included.
Of charges that have been disposed, the cumulative share reaching disposition within 90 days, 6 months, 1 year, and 2 years of filing. Each interval is cumulative — “within 1 year” includes everything within 6 months.
Based only on charges that have actually been disposed; pending charges are reported separately (below). A charge disposed the same day it was filed counts as within standard.
Counted at the case level: a case is “pending” if it has at least one charge with no recorded disposition — i.e. it has not yet fully resolved. Reported by filing year.
This is the active backlog behind the blank-disposition figure in the data-completeness panel: most blank dispositions are simply cases that have not closed yet, not missing data.
Pending cases grouped by how long they have been open, measured from filing date to today (0–90 days, 91–180, 181–365, 1–2 years, over 2 years).
Parish records in this dataset begin early 2023, so the oldest bucket is sparse. A very old case with no recorded disposition may reflect either a genuinely stalled case or a disposition that was never entered into the record — the oldest bucket should be read with that ambiguity in mind.
A case is treated as permanently closed only when every charge carries a clean-close disposition — dismissals, refusals, no-bills, paid-out, and a specific set of terminal outcomes.
Convictions are not treated as permanently closed, because a guilty plea or verdict can be followed by probation, suspended sentences, or jail terms that generate later court activity. Convicted cases continue to be revisited for updates.
For each presiding judge, the median and average days from filing to disposition, with a per-case-class breakout, shown only for judges with at least 20 disposed cases so the medians are stable.
This describes caseload processing time, not the performance of any judge. Judges hear different mixes of cases, the same case can have more than one judge associated with it, and judge attribution is only available for part of the caseload (see limitations). It should not be read as a ranking.
Warrants are reported two different ways, because “how often does failure-to-appear happen” and “who has a warrant right now” are different questions.
Failure-to-appear (FTA) rate — the share of cases filed in a year for which a warrant was issued at any point. This is a historical-incidence rate: it measures how often failure-to-appear occurred, and it does not decrease when a warrant is later recalled. For parish cases this is identified from warrant events in the court’s own case history.
Currently outstanding — cases whose warrant is active right now. For parish cases this is determined from the court’s own most recent structured warrant status (“outstanding” vs. “recalled”): a case counts only when its latest status is outstanding and no later recall is on record. This is the same definition used by the public warrants list, so the list and this figure agree. It deliberately does not use the raw warrant count shown on a case, because that count is cumulative — it rises when a warrant is issued but is never reduced when one is recalled, so it overstates who is currently wanted.
City Court warrants are identified from the current case status recorded by the City Court (Bench Warrant, Failure to Appear, or Regular Warrant). City Court status reflects the court’s present state and is refreshed continuously, so the city figure is a current-status measure.
Warrant information is provided to ground the published statistics and for reference only; it is not an official warrant registry. A warrant may be served or recalled after our most recent record of it, so an entry can lag the court. Each parish warrant on the list shows the date of the court status it is based on. For official warrant verification, contact the Lafayette Parish Sheriff’s Office or the Lafayette City Court.
Defendants appearing in two or more records, combined across Parish Court and City Court, matched by name and date of birth.
Name-and-birthdate matching is imperfect: some counts may combine, or fail to combine, people in edge cases. Treat the distribution as indicative, not exact.
Median, average, minimum, and maximum bail amounts from the case_bail_summary table, which is populated from bond-related fields in scanned case documents. Reported by case type and broken out by defendant race and sex where those fields are present in the case record. The median is reported first and should be preferred for comparison: bail amounts are heavily right-skewed, so a single very large bond can pull a group's average far above what is typical.
For the race breakdown, raw court-entered race codes are consolidated into four buckets (White, Black, Hispanic, Other) to correct for varied data entry, and the grouping is done before any statistic is computed so that each group's median is a true median. Any race or sex group with fewer than 5 cases is suppressed.
Bond amounts reflect the bail set at the case level as recorded in the document text; they may differ from the total bail ordered by a judge if multiple charges carry separate bond requirements. Only cases with a positive bond amount recorded are included — and because the figure exists only where a document was collected and read successfully, the covered subset may not be representative of all cases. Race and sex are court-recorded demographic fields from the search_results table and may be incomplete or inconsistently entered. These breakdowns do not control for charge severity, criminal history, or any other factor — a difference between groups here is a starting point for questions, not a finding of cause.
Median and average sentence length in days, by disposition, parsed from the raw sentence-length text recorded on charges. Because the raw text mixes units (“5 YEARS,” “180 DAYS”), each value is converted to days using its stated unit (years × 365, months × 30, weeks × 7). Values whose unit cannot be recognized are excluded rather than guessed, so the counts reflect only sentences that could be parsed with a known unit. Minimum 5 parseable charges per disposition to appear.
The conversion uses calendar approximations (a “month” is counted as 30 days) and reflects the sentence text as recorded, which may combine or omit concurrent terms, suspensions, and credit for time served. Treat these figures as indicative of scale, not as a legal computation of any sentence.
Each charge's Louisiana Revised Statute number is normalized (title and section, stripping data-entry suffixes) and mapped through a fixed, published crosswalk to one of eight offense categories — violent, property, drug, DUI, criminal traffic, non-criminal traffic, public order, and family/support. These categories align with the offense taxonomy used by national court-data projects such as Measures for Justice, which is itself modified from the offense codes the Bureau of Justice Statistics uses in the National Corrections Reporting Program (NCRP). Weapons offenses are classified under public order, per that convention; home invasion is classified as violent, as an enumerated crime of violence under La. R.S. 14:2(B).
Statutes not in the crosswalk are shown as Unclassified, never guessed. Because the court updates a charge's statute number when the charge is amended, categorization reflects the charge as finally recorded, not as originally filed. Rows in the source data that are hearings, orders, or writs entered as charges are classified as Court Event and excluded from the headline charge count.
This table counts charges. Projects that classify each case by its single most serious charge will produce different totals from the same underlying records, so these figures are not directly comparable to case-level offense breakdowns published elsewhere.
For charges with a guilty outcome or a deferred adjudication under Arts. 894/892.1 (which carry a sentence structure until any set-aside), the recorded sentence, suspended, and probation terms are parsed with a unit-aware, compound-capable parser (“7 years, 6 months” is supported) and classified: custodial only, custodial with probation, split (partly suspended), fully suspended, probation only, or no recorded terms. “Median net time” is the sentence minus its suspended portion — the custodial exposure the record shows.
Three kinds of rows are excluded and counted rather than averaged in: durations whose unit cannot be recognized; inconsistent rows where a suspension has no sentence or exceeds it; and probation terms over 8 years, which exceed any lawful term (La. C.Cr.P. arts. 893/894; the 8-year maximum applies to specialty-court probation) and are treated as data errors. Net time is the recorded term, not time actually served — credit for time served, parole, and concurrency are not visible in these fields. “No recorded terms” is typically a fine-only sentence, but parish court fine amounts are not captured, so the two cannot be distinguished.
Dispositions that resolve a charge outside straight conviction or dismissal, categorized by exact match against the court's recorded disposition vocabulary: misdemeanor deferrals under La. C.Cr.P. art. 894 (which, on successful completion, end in the conviction being set aside with “the same effect as an acquittal”), traffic deferrals under art. 892.1 (driver-improvement program), deferrals whose article is not recorded, specialty courts (R.S. 13:5301 et seq.), the District Attorney's pretrial diversion program (City Court “PTD” codes), screening outcomes (no true bill, refusal, quash), and competency/insanity findings.
Disposition codes not in the map are never guessed into a category. Two known asymmetries: felony deferrals under art. 893 are not marked in Parish Court records — a completed 893 appears only as a dismissal — so dismissal counts include an unknowable number of successful felony deferrals; and City Court “pending” deferral labels are not reliably updated on completion, so completed deferrals are undercounted and pending counts overstate active deferrals. A set-aside under art. 894 may still serve as a prior offense for enhancement (art. 894(B)(2)).
Three related measures. Convictions via amended or lesser charge: of charges resolving in a guilty outcome, the share recorded as an amended or lesser charge, by disposition year. Charge amendment pairs: the most common original→amended charge combinations — in Parish Court extracted from the “[AMENDED TO]” annotation the court embeds in charge descriptions; in City Court from the recorded original and amended violation fields. Mixed outcomes in multi-charge cases: multi-charge cases resolving with at least one guilty outcome and at least one dismissed or nolle prossed charge — the pattern consistent with count bargaining.
The record shows that a charge was reduced, not why; these figures measure the court record's account of plea practice, not prosecutorial intent, and cannot distinguish a negotiated dismissal from one made for evidentiary reasons. Reductions made before filing are invisible to court records entirely. Parish Court charges with the “APG” disposition code are excluded from the amended-plea measure pending public-records confirmation of the code's meaning. Amendment pairs carry a minimum-occurrence floor (5 in Parish Court, 10 in City Court).
Age at case filing, computed from the court-recorded date of birth, in the age bands used by comparable court-data projects (18–25, 26–35, 36–45, 46–65, over 65).
Records with ages under 18 or outside a plausible range are excluded and counted: they are a mix of juveniles charged as adults and date-of-birth entry errors, and the record cannot distinguish the two.
Felony and misdemeanor cases with an affirmative Indigent Defenders' Office marker in the docket activity record — an IDO appointment filing or an IDO-filed pleading — as a share of cases filed that year. The marker is taken from structured docket codes only.
Prose minute entries are deliberately not used: they over-match (they include referrals to the IDO that may not become appointments, entries stating the accused was not represented, and incidental mentions in service returns). The measure counts an appointment at any point in the case — counsel may later be retained privately — so it reflects “received an indigent-defense appointment,” not “represented by a public defender at disposition.” In the 15th JDC, indigent defense is provided by a public/private cooperative, so attorney names on these cases include private contract counsel; the docket marker, not the attorney's office, is what identifies the case as indigent defense. Because appointments enter the record at or after arraignment, the most recent filing years are incomplete by nature and their rates rise as cases reach arraignment.
Average days from the offense date on a charge to the case filing date, reported by filing year and case type. Only charges with both dates present and where filing follows the offense are included.
For City Court cases: average days from date of offense to the first scheduled arraignment (from case_events), and from the latest arraignment to the sentence date recorded on the case. Split by case type prefix — CC (criminal), CT (traffic), SC (small claims), ST (small claims traffic), DT (drug traffic).
Only cases with all required dates present and in the expected order are included. Case type is derived from the alphabetic prefix of the City Court case number.
No court dataset is complete, and pretending otherwise is the fastest way to mislead. Rather than a vague “some data may be missing” disclaimer, each statistics page shows a Data Completeness panel that quantifies exactly how much of each required field is missing, as a percentage. The principle throughout is simple:
A few consequences worth understanding:
Names in court records are entered by many hands over many years, so the same person often appears under several spellings. How the project consolidates those names is itself a source of potential error, so the rules are deliberately conservative and are described here.
Judge names are read primarily from the structured “Judge: … Division: …” entries in the record (the reliable source), with looser narrative mentions used only as a fallback. Spelling variants are then consolidated to a canonical name using a reviewed mapping built against the court's own published roster of judges and divisions. Because the dataset spans several years, it includes judges who have since rotated off or retired; those historical judges are kept, not dropped. Judge attribution is available for roughly four-fifths of cases — where a judge could not be identified from the record, the case is simply absent from per-judge figures.
Defense attorneys are consolidated only on strong, unambiguous signals (such as a shared contact address), never by fuzzy name-similarity, because two different attorneys can have similar names. As a result, the attorney list may show the same person under more than one spelling rather than risk merging two different people. For prosecutors (DA/ADA), where the records carry no such strong signal, names are normalized for spacing and punctuation, and likely duplicates — such as a name entered in reversed order — are flagged for the reader rather than automatically combined. A wrong merge would silently misattribute one person's case outcomes to another, so the project errs toward showing a visible possible-duplicate note instead.
Officer names are extracted automatically, using optical character recognition (OCR), from the text of scanned court documents (affidavits, bond attachments, incident reports, traffic summonses, and bills of information). This is a sparse signal: only a small fraction of cases contain a document a name can be read from, so most cases have no extracted officer at all, and the absence of an officer means nothing. The result is not an official roster of any agency.
For each officer the project shows the document their name was extracted from rather than asserting a role such as “arresting officer.” The underlying documents do not reliably state a role, so being named in a document means only that the name appeared there — not that the officer made an arrest or did anything wrong. Names are cleaned for display by removing rank and title words (for example, “Detective Det.” or doubled titles introduced by OCR), leaving the plain name.
Officers are consolidated by badge number where one is present (badges appear only on certain structured document types, and where present are clean and reliable), and otherwise by a normalized name within the same agency. Agency-name variants (for example, “Lafayette Police Department,” “Lafayette City Police Department,” and abbreviated forms) are consolidated to a canonical agency using a reviewed mapping, so an officer is not split across spellings of their own department. As with attorneys, similar names are never merged by similarity alone — spelling variants of a person’s name are left as separate entries rather than risk combining two different people.
In the interest of honesty, here are the things this dataset does not do well, or cannot do:
The heatmap geocoding coverage is also ongoing: records with usable addresses that have not yet been geocoded are retried continuously, so the proportion of mapped points improves over time. The coverage breakdown on the Heatmap tab shows the current state per year.
If you spot an error, want to understand a number, are a journalist, researcher, or community advocate who could use the underlying records, or represent a court office with a correction, please reach out:
Corrections about specific records are taken seriously — the authoritative version always lives with the court, and this project aims to reflect that record as faithfully as the available data allows.