Sorting Custody Credits Between Cases: The Day That Counts Twice, and the Day That Earns Nothing

The verdict came back guilty. The fight over the sentence is mostly done. What is left is a number, and it is the number that decides when you walk out: how many of the days you already sat in jail get taken off the end. If you were picked up on more than one case, the judge does not add up every day you were in the building and hand you the total. Each day gets sorted, case by case, before anything is subtracted.

Two questions get asked, in this order. Which days belong to this case. Then what those days are worth. People lose credit at the first question and never find out, because the second question is the only one anybody explains.

Question one: which days belong to this case

The rule is one sentence of Penal Code section 2900.5, subdivision (b): “credit shall be given only where the custody to be credited is attributable to proceedings related to the same conduct for which the defendant has been convicted.”

Attributable is the whole fight. The test the California Supreme Court uses is a but-for question, and you can ask it about any single day. If this case did not exist, would you have gone home that day? If the answer is yes, the day belongs to this case. If you would have stayed in jail anyway, on a parole hold or a different case or a warrant out of another county, the day does not belong here.

In re Joyner (1989) 48 Cal.3d 487 put it this way: time already credited against a sentence for unrelated offenses cannot be treated as attributable to a later case “unless it is demonstrated that the claimant would have been at liberty during the period were it not for a restraint relating to the proceedings resulting in the later sentence.” People v. Bruner (1995) 9 Cal.4th 1178 tightened it. Showing that this case was a reason you were in custody is not enough. You have to show it was a but-for reason.

There is a large exception, and it gets missed. The strict but-for test exists to stop double credit. Where double credit is impossible, the test does not apply. In re Marquez (2003) 30 Cal.4th 14 involved a defendant whose other case had been dismissed, so the choice was between counting the days once or counting them never. The Supreme Court held the strict causation rule had no work to do. If the other case went away, say so, and say that the choice in front of the court is credit once or credit not at all.

One day can belong to two cases at the same time

This is where people assume the worst and give up days they are entitled to. A day is not a coin that can only land in one pile.

People v. Jacobs (2013) 220 Cal.App.4th 67 says that if you were in custody for even part of a day in each of two cases on the same date, you get a full day of credit in both. And on June 25, 2026, the California Supreme Court settled the larger version of the question in People v. Cofer (2026) 20 Cal.5th 1.

Cofer holds that “proceedings” in section 2900.5(b) means an individual criminal case. Resolving five cases in one plea deal at one hearing does not melt them into a single proceeding. Where the sentences are concurrent, credit for a single stretch of custody is given against each of those concurrent sentences, so long as the custody is attributable to each of those cases. The Court disapproved People v. Kunath (2012) 203 Cal.App.4th 906 to the extent it said otherwise.

Read the second half of that sentence as carefully as the first. Cofer does not say every case gets every day. The Supreme Court reversed the Court of Appeal precisely because the trial court had awarded credits in each case for the time spent in custody in that case, rather than giving every case the total from all of them. The sort still happens. It just no longer stops a day from landing in two places when it honestly belongs in both.

Question two: what the surviving days are worth, and where the odd day dies

Actual days are only half of it. The other half is conduct credit, which is what most people mean when they say good time or work time. It comes from Penal Code section 4019, and since realignment the rate is two for two.

Subdivisions (b) and (c) each give one day off for every four-day period of confinement, one for work and one for behavior. Subdivision (f) states the result plainly: “It is the intent of the Legislature that if all days are earned under this section, a term of four days will be deemed to have been served for every two days spent in actual custody.”

Two actual days become four. Which means conduct credit is earned in pairs, and a day with no partner earns nothing.

People v. Chilelli (2014) 225 Cal.App.4th 581 says it without hedging: “under the two days for every two days rate of accrual, there can be only an even number of presentence conduct credits,” because “conduct credits are given in two day increments and no rounding up is permitted.” A defendant with 25 actual days gets 24 conduct days. People v. Whitaker (2015) 238 Cal.App.4th 1354 held the same for 327 actual days, which produced 326.

So the direct answer to the question people actually ask: one day of custody, standing by itself, earns one actual day and zero conduct days. There is nothing to pair it with. The arithmetic does not round in your favor, and no judge has discretion to make it round.

The move that saves the odd day

That is only true if the day really is standing by itself. It usually is not.

People v. Taft (2026) 119 Cal.App.5th 380, decided March 20, 2026, holds that “a defendant is entitled to have time spent in noncontinuous custody aggregated for the purpose of calculating [conduct] credits.” Separate stretches get added together first. The formula runs once, on the total.

That changes the answer for anyone arrested more than once on the same case. One day at the first arrest and one day at the second are two separate periods of one day each. Computed separately, each earns zero. Added together, they are two actual days, which is a complete pair, and a complete pair earns two.

Probation reports frequently list custody stretch by stretch and compute conduct credit on each line. On any set of stretches with more than one odd number in it, that method produces a total that is short. Add the actual days first, then apply the formula once, then compare the two numbers before the judge signs anything.

Three ways the credit still comes out short

The four-day floor

Section 4019, subdivision (e) reads: “A deduction shall not be made under this section unless the person is committed for a period of four days or longer.” This is misread constantly as a rule about how long you were in jail. It is not. People v. Dieck (2009) 46 Cal.4th 934 construed the equivalent older threshold and held that the subdivision “sets forth a minimum duration of ordered commitment, not a minimum term of presentence incarceration.” The word in the statute is committed, not confined. What matters is the length of the sentence the court orders, not the number of days you actually sat. Five days of presentence custody on a one-year commitment clears the floor easily.

The 15 percent cap

Penal Code section 2933.1 applies to anyone convicted of a felony listed in section 667.5, subdivision (c), the violent felony list. Subdivision (c) overrides section 4019 outright: the maximum credit against county jail time following arrest “shall not exceed 15 percent of the actual period of confinement.” People v. Ramos (1996) 50 Cal.App.4th 810 and People v. Nunez (2008) 167 Cal.App.4th 761 apply the cap across the whole sentence when any one count is a violent felony, including the terms for counts that are not. People v. Rosales (2013) 222 Cal.App.4th 1254 confirmed realignment did not repeal it. On a short stretch, 15 percent of a handful of days rounds down to nothing, which is one reason a credit calculation can come back with an actual number and a zero next to it.

Murder

Section 2933.2, subdivision (c) is absolute for anyone convicted of murder under section 187: “no credit pursuant to Section 4019 may be earned against a period of confinement in, or commitment to, a county jail.” Actual days still count. Conduct days do not exist. The section applies to murders committed on or after the date it became operative.

What to have ready before the judge does the math

The calculation happens fast, often at the end of a long hearing, and section 2900.5, subdivision (d) makes it the court’s duty to fix the dates and the total and put the total in the abstract of judgment. A number that goes in wrong follows the case. Bring the material that lets the court get it right the first time.

1. Every booking and release date on every case, taken from the jail records, not from memory.
2. Which case or cases each stretch of custody was actually held on, taken from the booking sheet, not the charging document.
3. Whether bail was ever set on this case, at what amount, and whether it was reachable.
4. Whether a hold from another case, a parole or probation revocation, or an out-of-county warrant was the reason you stayed in.
5. Whether any of these days have already been credited against a sentence in another case, and the paperwork showing it.
6. Whether any other case has been dismissed, which puts you inside Marquez instead of Bruner.
7. Your own total, computed by adding all the actual days first and applying the two-for-two formula once, ready to put against the probation department’s number.

Read more: the but-for showing, and the record that actually carries it

The reason so much presentence credit disappears is that the burden sits on the person least equipped to carry it, and the showing has to be evidentiary.

What Bruner requires. The defendant in Bruner was in custody on a parole revocation that rested on several grounds at once, one of which was the cocaine possession that became his new case. He had already received full credit for that custody in the revocation proceeding. The Supreme Court held he could not “prevail simply by demonstrating that the misconduct which led to his conviction and sentence was ‘a’ basis for the revocation matter as well.” He had to show “that but for the cocaine possession leading to his current sentence, he would have been free, or at least bailable, during that presentence period.” He could not, because the revocation stood on its own without the new offense, so the duplicative credit was denied.

What Joyner requires. Joyner involved a California hold lodged against a defendant already in custody in Florida. The Court noted there was “no evidence that petitioner ever posted bail on the Florida charges or that he could have obtained release had the California hold not been placed against him.” Absence of evidence decided it. The hold’s existence proved nothing about whether it was doing the work of keeping him in.

So the showing is about release, not about charges. The question is never whether this case contributed to your custody. It is whether removing this case from the world would have put you on the street. Three pieces of proof do most of the work:

First, bail. If bail on this case was set at an amount within reach and the other matter carried a no-bail hold, the hold is the reason for the custody and this case is not. That cuts against you. Reverse the facts and it cuts for you. Either way it needs to come from the record: the bail amount, the hold, the date each was placed. Our guide on conditions of release after arraignment covers how those holds get placed and lifted.

Second, the booking records. Jails record which case each stretch of custody is held on and when a hold is added or dropped. That document is what turns an argument into evidence, and it is the thing most often missing at sentencing.

Third, testimony or a declaration where the money mattered. Odom is the warning. Counsel’s statement that the client would have posted bail was not evidence and was treated as nothing. Someone who would have posted has to say so under oath, and be able to say with what.

The Marquez off-ramp, and when to reach for it. Strict causation exists to prevent one period of custody from producing two sentences’ worth of credit. When the other case has been dismissed, or the other sentence never happened, there is no second sentence and no windfall to prevent. Marquez makes that structural rather than discretionary. The framing matters at the podium: tell the court the choice in front of it is credit once or credit never, and that the rule it is being asked to apply was built for a different choice.

Where Cofer changes the argument. Before June 2026 there was real authority for the position that cases resolved at one hearing became one proceeding, which pushed toward one shared pot of credit. Cofer rejects that reading and disapproves Kunath to the extent it supported it. The practical effect runs both directions. Each case now gets what is attributable to it, which means a day held on all three cases counts in all three concurrent sentences. It also means you cannot ask for the aggregate across cases and expect it to survive. Come with the attribution done case by case.

Read more: the arithmetic, worked out

Actual days and conduct days are two separate columns. Add them at the end, never in the middle.

Actual days on this case	Conduct days under section 4019	Total credit
1	0	1
2	2	4
3	2	5
25	24	49
143	142	285
327	326	653

The rule the table encodes. Take the actual days. If the number is even, the conduct days equal it. If the number is odd, subtract one. Add the two columns. That is the whole calculation at the full rate, and it is what Chilelli and Whitaker hold.

Where the aggregation rule bites. Suppose three separate stays on the same case: 7 days, 5 days, and 1 day. Computed stretch by stretch, the conduct credits are 6, 4, and 0, for 10. Aggregated as Taft requires, the actual total is 13, the conduct credit is 12, and you are two days better off. The gap grows with the number of odd stretches. Anyone arrested repeatedly on the same case, which describes most probation cases and a good share of domestic violence cases, is exposed to this.

Where the 15 percent cap bites. Under section 2933.1 the calculation is not two for two at all. It is 15 percent of the actual period of confinement, computed and then truncated. Thirty actual days produce 4 conduct days, not 30. Six actual days produce 0. The cap reaches every count in the case once any single count is a violent felony under section 667.5, subdivision (c), which is why a plea that leaves one violent count standing changes the credit math on everything else in the file.

A caution on section 4019 itself. There are two versions of section 4019 on the books. The one quoted here is the version currently in force, effective January 1, 2024, and it repeals itself on January 1, 2028, with a successor already enacted. Any calculation for custody served in a different window should be run against the version operative then, because the rate has changed more than once since 2010.

For what happens to the days after the credits are set, see our guides on alternatives to custody and how long probation lasts.

The short version

Sort first, then multiply. A day counts toward this case only if this case is why you were not free, and proving that is your job. A day can count in two cases at once when the sentences are concurrent and the day honestly belongs to both, which is what Cofer settled in June. Conduct credit comes in pairs, so a single leftover day earns nothing on its own. And a leftover day is only leftover if nobody added your custody stretches together first, which under Taft they are required to do.

Last reviewed: September 2026. This is general information about California law, not legal advice about your case. Credit calculations turn on booking records and on the exact charges, and the version of section 4019 that applies depends on when the custody was served.