California Bail Statistics: Jail and Pretrial Release Data

Updated September 27, 2026. Figures below were checked against the Judicial Council's 2026 report and current Board of State and Community Corrections data. Reporting dates and coverage are identified with each table. This is general information, not a prediction or legal assessment of an individual case.

What do the latest court-release figures show?

The Judicial Council's Pretrial Services Program Year 4 report, published April 24, 2026, covers activities from July 1, 2024 through June 30, 2025. Its program-data section provides a useful starting point for understanding recorded bookings and releases.

Coverage matters: Footnote 24 states that Year 4 data do not include the Los Angeles court while it undergoes a countywide case-management transition. Treat the following figures as results for the report's participating data set, not a complete count of every California booking or a national estimate.

Source: Judicial Council, April 2026 report, printed page 9. The figures cover July 2024 through June 2025 and exclude Los Angeles court data. The report uses booking events, which should not be assumed to represent unique people.

A recorded release does not always mean someone went home on bail

The report separates pretrial release from outcomes such as transfer to another agency, no charges filed or charges dismissed, time served or transfer to state prison, and other recorded resolutions. A transfer can move someone from one form of custody to another. Counting it as a recorded release from the original booking does not mean the person returned home.

This distinction prevents a common error: calling the overall release percentage a "bail success rate." A family researching bail needs the category that actually describes release on financial conditions. Someone researching case processing may need the wider collection of booking resolutions. Both can be useful if the label matches the question.

When you see a claim that a certain percentage of people "got out," ask what that phrase includes. Does it mean release pending trial, transfer to a different institution, completion of a sentence, dismissal, or every recorded exit combined? The answer determines whether the statistic applies to the issue you are trying to understand.

The report's publication year is different from its data year

A report released in 2026 can describe events in an earlier fiscal year. A fiscal year is a reporting period that may cross two calendar years. Here, the period begins in July 2024 and ends in June 2025. The April 2026 publication date tells you when the report became available, not when each booking happened.

Keep both dates when sharing a figure. For example, describe it as "the April 2026 report's July 2024 through June 2025 data." That is more informative than calling it a current daily jail count. It also helps readers understand why a newer local dashboard might show a different population or use a later month.

How many releases were recorded as bail?

The same Judicial Council report divides pretrial release into four categories and separates releases within two days of booking from those occurring later. The table below uses the exact counts in the report's text, rather than the rounded labels in its charts.

Source: Judicial Council Year 4 report, printed pages 10 and 11. Period: July 2024 through June 2025. Los Angeles court data are excluded. The category totals should not be described as a count of unique people statewide.

Adding the report's two bail counts gives 82,389 recorded bail releases in the covered data. That is an author calculation: 70,038 plus 12,351. Adding the four categories across both timing groups gives 281,223 pretrial releases, another calculation from the reported values.

What the categories mean for a family

In this report, bail means release on secured financial conditions. The category does not isolate commercial surety bonds from every other financial form of bail. It is therefore not a count of bail-company customers, bond contracts, or Bail Hotline clients.

Jail cite and release describes release with a promise to appear. Own-recognizance release, often shortened to OR, is a judicial release based on a promise to appear, as described in this report. Pretrial monitoring adds monitoring conditions and the obligation to comply with them. Those labels describe different recorded routes, not a menu that every person can select without a court or agency decision.

For an actual case, ask which route has been authorized and what written conditions accompany it. A family might hear that someone is being "released" without knowing whether the person must report to pretrial services, follow an order limiting contact, or appear at a particular courthouse. The paperwork and current orders supply that missing detail.

The report's two-day division is also a measurement choice. It helps describe when recorded releases occurred. It does not guarantee that a new defendant will leave custody within that period, establish a universal processing time, or replace the legal rules governing court appearances.

Why does the denominator change the answer?

The denominator is the group at the bottom of a fraction. In a percentage, it tells you "out of what?" Changing that group changes the meaning of the result, even when the number at the top stays the same.

Using the Judicial Council's displayed totals, these two calculations answer different questions:

• 266,327 divided by 403,339 is about 66%. This compares releases within two days with all reported releases in the covered period.
• 266,327 divided by 521,564 is about 51%. This compares that same count with the reported new-offense booking total. This second percentage is an author calculation, not the report's stated share of releases.

Neither calculation should be presented as the percentage of all Californians who posted a commercial bail bond. The numerator includes more than bail, the reporting population has a coverage limit, and booking events are not necessarily unique people. The calculations are descriptions of the published totals, not individualized probabilities.

Booking events and people are different units

A person can have more than one booking event. For that reason, a table counting events may be useful for measuring workload without telling you how many distinct people experienced that workload. A count of criminal cases is another unit: one person's legal situation can involve multiple charges, cases, or custody events.

Hypothetical example: Imagine that a small reporting system records 100 booking events, and some people were booked more than once. The system has not necessarily recorded 100 different people. You would need a method for identifying repeat people before producing a unique-person count. This example explains the measurement issue; it is not an estimate of repeat bookings in California.

The same care applies to percentages. "People with at least one missed hearing," "hearings missed," and "cases closed unsuccessfully" can produce different results because they count different things. Before comparing two studies, write down each study's unit of analysis in ordinary language.

Corrections to earlier reports affect comparisons

Appendix D of the April 2026 report corrects several earlier figures. It explains that earlier release-rate calculations used total bookings, including events connected with prior arrests. The updated approach uses new-offense bookings. The appendix also corrects the period stated in the 2025 report's executive summary and updates previously reported demographic figures.

That means an older percentage should not automatically be placed beside a new one and described as improvement or decline. First check whether the reporting group, definition, time period, and coverage are comparable. For the latest report, the Los Angeles exclusion is a separate reason to avoid a casual statewide year-to-year comparison.

If you are writing a presentation, school paper, or news item, use the corrected values and include the appendix reference in your source notes. A publication date alone does not tell you whether the number has since been revised.

What do current local jail populations show?

The Board of State and Community Corrections, or BSCC, operates the Jail Profile Survey. Its current public query includes data through the second quarter of 2026. This is a different collection from the Judicial Council's court-release report.

The following four agency snapshots come from the official JPS query, using June 2026 and the fields for total average daily population and non-sentenced male and female average daily population. They are examples from selected reporting agencies, not a statewide total or a ranking of county performance.

Source: BSCC public query, checked September 27, 2026. The non-sentenced column is calculated by adding each agency's reported male and female non-sentenced fields. The source values are rounded monthly averages. No missing-value estimate or statewide aggregation was used for this table.

Los Angeles appears here because this table uses BSCC agency submissions for June 2026. That does not fill the Los Angeles gap in the separate court-release data for July 2024 through June 2025. The two collections measure different activity during different periods; they cannot simply be added together or used as each other's denominator.

Average daily population is not monthly bookings

Average daily population, or ADP, describes how many people were housed on an average day during the reporting period. A booking count describes entries into custody. A facility can have many short stays, fewer long stays, or a mixture. The same average population can arise from different patterns of admissions and length of stay.

Hypothetical example: A 30-day month with a total of 300 occupied-person days produces an ADP of 10. That could represent 10 people each staying all 30 days, or many different people staying for shorter periods that total 300 days. The example illustrates the calculation; it does not describe any agency in the table.

The BSCC survey workbook explains what the population fields include and exclude. They are not an unrestricted count of everyone who passed through a holding or booking area. Those reporting boundaries matter when comparing a jail dashboard with an arrest count or a news report about daily admissions.

Non-sentenced does not mean detained only because bail was unaffordable

BSCC defines the non-sentenced category to include people who have not been sentenced on all pending charges. A person who has been sentenced on some charges but still has another charge pending can be included. The category is therefore not the same as a count of people with no convictions, people eligible for immediate release, or people detained solely because of money.

To answer a question about why a particular person remains in custody, you need the person's actual case and custody information. The aggregate label does not identify the controlling order, another hold, a pending charge, or the person's financial situation. Avoid turning a population category into an unsupported explanation of every person's circumstances.

What do the numbers say about missed court and rearrest?

Missed appearances, new arrests, new convictions, and unsuccessful program closures are different outcomes. A source must define the outcome before its percentage can be interpreted. It also needs to identify the people or events counted and the period during which outcomes were tracked.

For example, the Judicial Council report's monitoring section defines a successful closure as case resolution or disposition, or a change to unmonitored release. Its unsuccessful category can include a return to custody, a new crime, or failure to appear. It also has an other/unknown category. Those definitions do not turn a closure percentage into a statewide commercial-bail rearrest or court-appearance rate. See the report's monitoring discussion, printed page 12.

When evaluating a claim about court attendance, ask whether it counts any missed hearing, a warrant, a failure that remained unresolved, or something else. For a new-offense claim, ask whether the measure is an arrest, a charge, or a conviction. Those events occur at different stages and should retain their actual labels.

Also ask how long each person was observed. A group followed for a few weeks and a group followed for a year have different opportunities for a recorded event. A responsible comparison explains those differences instead of putting two percentages beside each other without context.

Can these statistics prove that a bail policy changed crime?

The tables in this guide describe reported activity. On their own, they do not establish that a particular release policy caused an increase or decrease in crime. A meaningful evaluation needs a clearly defined policy, a relevant time period, appropriate comparison groups, and attention to other changes occurring at the same time.

BSCC's guidance on interpreting trends identifies factors such as local resources, jail-unit openings or closures, population characteristics, and economic conditions. It cautions against drawing conclusions about comparative county performance from trends alone.

For a reader, the practical test is simple: does the source measure the policy effect it claims, or does it merely show that two things happened during the same period? A change in a jail count can raise a useful question without answering why the count changed. Similarly, a study from a different state should not silently become a California statistic.

How do these statistics connect to an actual bail decision?

Statistics help describe the system. An individual bail decision depends on the applicable law, current court orders, and case-specific information. A reported average or percentage cannot tell a family the amount currently required in one person's case.

Penal Code section 1269b addresses acceptance of bail and the role of a warrant, county schedule, or judicial order. It also requires countywide schedules to be revised annually. A historical average is therefore a poor substitute for checking the current amount and controlling order.

Penal Code section 1275 directs judges to consider public protection, the seriousness of the charge, the defendant's prior criminal record, and the probability of appearing in court, with public safety as the primary consideration. These are individualized issues. They are not answered by the release rate for everyone in a large data set.

In In re Humphrey, decided in 2021, the California Supreme Court held that a person cannot be detained solely because of an inability to afford bail. The decision requires consideration of ability to pay when financial conditions are necessary and addresses nonfinancial alternatives and the findings required for lawful detention. It does not make a statistical average the correct amount for an individual defendant or guarantee release in every case.

For a fuller explanation of scheduled amounts and case-specific issues, see our California bail-amount guide. Ask the defense lawyer about legal options to review an amount or conditions, and confirm current custody information before making plans around an expected release.

Keep the bail amount and the bond agreement separate

A court's bail amount and the terms of a bond agreement answer different questions. When speaking with an agent, ask for an explanation of the proposed agreement, payment obligations, any collateral requirements, and the cosigner's responsibilities. Use the actual written terms rather than a percentage repeated in an old statistics article.

It helps to write two separate headings in your notes: "Court and custody requirements" and "Bond agreement." Under the first, record the case, amount, conditions, and next appearance. Under the second, record the questions you need answered before signing. This keeps a general internet number from being mistaken for the terms of a particular transaction.

What should a family check after an arrest?

If the immediate concern is helping someone in custody, gathering the right facts will usually be more useful than searching for a statewide average. A short, organized record also reduces confusion when several family members are making calls.

1. Confirm the person and facility. Use the person's correct name and booking identifier when available. Distinguish the agency that made the arrest from the facility currently holding the person.
2. Confirm the current case information. Ask which case or cases are involved, whether a bail amount is available, and whether another order or hold affects release. Do not assume that one listed amount resolves every custody issue.
3. Obtain the actual release instructions. A general label such as bail, OR, or monitoring does not replace the conditions. Record any required check-in, appearance, or restriction from the relevant paperwork.
4. Identify the next court event. Keep the court, case number, date, time, and appearance instructions together. If information conflicts, have counsel or the appropriate court office help resolve it.
5. Ask about the written bond terms. A prospective cosigner should understand the agreement and obligations before signing. Keep copies and record the contact for questions or changes in circumstances.
6. Plan for communication and transportation. Keep a working phone number, a practical pickup plan, and a way to reach the people involved. Statistical release intervals are not a promised pickup time.
7. Keep a dated call log. Note who provided information and when. Custody and court information can change, so a dated record is more useful than a number copied without context.

For someone collecting statistics instead of arranging bail, keep a different record: the source title, publication date, reporting period, jurisdiction, unit counted, definition, denominator, and any corrections. The two tasks require different information. A family needs the current case facts; a statistical comparison needs consistent measurement.

Frequently asked questions

Does the report's 77% release figure mean 77% posted bail?

No. The report's overall release count includes several categories, including outcomes other than pretrial release. Bail is one of the separately identified pretrial categories. The figure also belongs to a defined reporting period and excludes Los Angeles court data, so it should not be presented as a complete statewide bail-posting rate.

Do the latest court figures include Los Angeles?

The Year 4 report's program data exclude the Los Angeles court, as its footnote explains. The June 2026 Los Angeles Sheriff's Department snapshot in this guide comes from BSCC's separate Jail Profile Survey. One source's local jail population cannot be inserted into the other source's booking-and-release totals.

Is a booking count the number of different people arrested?

Not necessarily. A booking is an event, and a person can have more than one event. A unique-person count requires a method for identifying repeats. Keep the source's original unit instead of replacing "bookings" with "people" in a headline or summary.

Are all non-sentenced jail occupants unable to afford bail?

The BSCC category does not establish that. It includes people who have not been sentenced on all pending charges, including some with a sentence on one charge and another charge still pending. The aggregate field does not explain the controlling custody order or financial situation in each case.

Is there one California average bail amount I should use for my case?

An average cannot establish the amount in a particular case. Check the current booking and court information, including the controlling warrant, schedule, or judicial order. The datasets used here describe populations and release activity; they do not supply a personalized bail quote or bond agreement.

Does release within two days in a report promise a two-day release?

No. It is a reporting category describing recorded events. The actual case may involve different orders, conditions, processing steps, or other custody issues. Confirm the current situation with the appropriate professionals before treating any reported interval as a travel or pickup plan.

Is an unsuccessful monitoring case the same as a new arrest?

No. The report groups more than one kind of outcome under unsuccessful closure. A percentage using that category cannot be relabeled as a new-arrest, new-conviction, or failure-to-appear rate. Use the source's definition and identify the group followed before making a comparison.

What should I include when quoting a bail statistic?

Include the source, reporting dates, geographic coverage, unit counted, and denominator for a percentage. Identify any calculation you performed and check for later corrections. For the figures here, the Los Angeles coverage limit and the distinction between release events and unique people are essential context.

Sources and update notes

The principal numerical sources are the Judicial Council's April 2026 Pretrial Services Program report, including its correction appendix, and the BSCC Jail Profile Survey query. The JPS workbook, dashboard guidance, and research guidance explain definitions, data revisions, and interpretation limits.

The BSCC database can be revised after agencies submit corrections. The figures in this guide are therefore a dated reading of the available sources. Future updates should recheck both the numbers and their definitions, while an individual family's next step should follow the current case information and orders.