Donohue v. AMN Services, LLC

(2021) 11 Cal.5th 58
Every short meal punch in the time-clock data now creates a rebuttable presumption that the employer failed to provide a compliant meal period. Rounding cannot be used to round away violations. The burden shifted.

Two holdings, each unanimous, each written by Justice Liu, with Court of Appeal Justice Hoffstadt sitting pro tem. First: employers cannot round time punches in the meal period context — the meal period provisions are designed to prevent even minor infringements, and rounding is incompatible with that objective. Second: time records showing missed, short, or delayed meal periods raise a rebuttable presumption of meal period violations, and the presumption applies on the merits — including at summary judgment — not merely at class certification.

The court adopted Justice Werdegar's Brinker discussion of the rebuttable presumption 'in full,' disapproved, to the extent inconsistent, Silva v. See's Candy Shops, Inc. (2016) 7 Cal.App.5th 235 (See's Candy II) and Serrano v. Aerotek, Inc. (2018) 21 Cal.App.5th 773 — both had refused to apply it at summary judgment — and confirmed the rebuttal paths: premiums paid, or employees genuinely relieved of duty who chose to work.

Kennedy Donohue worked as a nurse recruiter in the San Diego offices of AMN Services, LLC, a healthcare staffing company. AMN's written meal policy was Brinker-compliant on its face. The problem was the timekeeping layer beneath it: AMN's Team Time system rounded every punch to the nearest 10-minute increment. A lunch from 11:02 to 11:25 — 23 actual minutes — was recorded as a compliant half hour; a meal beginning five hours and five minutes into the shift was recorded as starting at exactly five hours.

The rounding did not just distort the records; it disabled the compliance machinery built on top of them. AMN's system paid an automatic premium (before September 2012) or triggered a dropdown menu asking the employee whether the noncompliant break was voluntary (after September 2012) — but only when the rounded records showed a violation. A meal period that was short or late in fact, and compliant only after rounding, never triggered either safeguard. Donohue's statistics expert counted 40,110 short and 6,651 delayed meal periods concealed this way, representing $802,077.08 in unpaid premiums. The trial court granted AMN summary judgment; the Court of Appeal affirmed, reasoning that neutral rounding was as lawful for meal periods as for work time. The Supreme Court reversed on both grounds and remanded for new summary adjudication motions.

Rounding entered California as a borrowed practice: 29 C.F.R. § 785.48, a 1961 federal enforcement regulation, tolerates timeclock rounding that averages out; the DLSE folded it into its enforcement manual; and See's Candy Shops, Inc. v. Superior Court (2012) 210 Cal.App.4th 889 (See's Candy I) adopted it for California wage calculation when neutral on its face and as applied. Donohue withheld any blessing — See's Candy I's validity remains undecided — and held that meal period rounding fails even the borrowed standard's own terms.

The meal period context is different in kind. Labor Code § 512 and the wage orders draw precise lines: 30 minutes, by the end of the fifth hour. The premium under § 226.7 is triggered by any shortfall, however small, because the statute compensates for the lost break, not for lost minutes of pay. A rounding system that treats a 23-minute lunch as a 30-minute lunch does not approximate compliance; it erases the violation.

The net-neutrality defense that carried the day below answered the wrong question: AMN's expert showed rounding overcompensated the class by 85 work hours, but overpayment for time worked says nothing about the premiums owed for noncompliant breaks. The flaw is asymmetry: rounding never triggers premiums that are not owed, yet 'it does not always trigger premium pay when such pay is owed.' The rounding analysis closes with a warning — as technology continues to evolve, the practical advantages of rounding may diminish further; AMN itself moved to an unrounded system after this lawsuit was filed.

The second holding converted a decade-old concurrence into binding law. In Brinker, Justice Werdegar wrote separately to explain that when an employer's own records show no meal period for a shift over five hours, the burden should shift: employee waiver 'is not an element that a plaintiff must disprove as part of the plaintiff's case-in-chief' but 'an affirmative defense' the employer must plead and prove. Donohue adopts that framework without qualification, and extends it expressly to short and delayed meal periods — not just missed ones — and to the summary judgment stage.

The rationale is the employer's record-keeping duty. The wage orders require accurate meal period records. If the records are accurate, the presumption costs a compliant employer nothing — the records reflect true liability. If the records are incomplete or ambiguous, the party who controlled the record-keeping bears the consequence: 'where the employer has failed to keep records required by statute, the consequences for such failure should fall on the employer, not the employee.' Anything else would reward employers for not recording breaks at all.

The doctrine's basis also fixes its boundary. The presumption 'derives from an employer's duty to maintain accurate records of meal periods' — and no parallel recording duty exists for rest breaks: the wage orders' records provision states that 'authorized rest periods need not be recorded,' which is why the experts in Alberts v. Aurora Behavioral Health Care (2015) 241 Cal.App.4th 388 modeled meal data only. Punch records therefore cannot generate a rest period presumption; rest claims require affirmative policy evidence and testimony. The limit is evidentiary, not substantive, but it strips the cheapest proof engine from half the break case.

Donohue is not a strict-liability regime, and the court said so directly, rejecting AMN's argument that the presumption creates 'automatic liability' or a duty to police breaks. Brinker's provide-not-ensure standard survives intact: the employer is liable only if it failed to provide the opportunity, and an employee's genuinely voluntary choice to shorten, delay, or skip a break creates no liability. The presumption is rebutted with evidence that premiums were paid or that employees were relieved of duty and chose to work — and the court reaffirmed that representative testimony, surveys, and statistical analysis are available tools for making that showing at scale.

The opinion even supplies the compliance blueprint. The presumption does not compel surveillance; it 'requires employers to give employees a mechanism for recording their meal periods' and to see the mechanism used properly. Team Time's dropdown — which asked the employee to attest, punch by punch, whether a short or late break was their own choice — 'would have ensured accurate tracking of meal period violations if it had simply omitted rounding.' Employers that build that apparatus convert the presumption from a threat into a records-based defense.

The question Donohue reserved — whether neutral rounding survives anywhere in California wage law — went up almost immediately. In Camp v. Home Depot U.S.A., Inc. (2022) 84 Cal.App.5th 638, the Sixth District read Donohue and Troester v. Starbucks Corp. (2018) 5 Cal.5th 829 to prohibit rounding of work time where the employer captured actual minutes worked. The Supreme Court granted review in February 2023 (S277518) on whether neutral time-rounding is permitted at all. Fully briefed since late 2023, the case still sat on the court's official pending-issues list in April 2026 with no argument date — and every rounding policy still in operation runs on borrowed time against that docket.

The presumption's sharpest published test so far, Allison v. Dignity Health (2025) 112 Cal.App.5th 192, came at certification rather than summary judgment: punch data showing unpaid noncompliant meal periods on more than 70 percent of shifts triggered the presumption and initially carried certification, yet the First District affirmed decertification because the employer's rebuttal evidence of idiosyncratic employee choice made the affirmative defense predominately individualized. The pointed corollary: contemporaneous records of the reasons for noncompliant punches might have made the defense triable classwide.

On the waiver side, Donohue distinguished the colloquial choice to work through a provided break from the statutory § 512(a) waiver for shifts of six hours or less. Bradsbery v. Vicar Operating, Inc. (2025) 110 Cal.App.5th 899 enforced the statutory kind — prospective, written, revocable waivers for five-to-six-hour shifts — a genuine safe harbor for short-shift workforces when revocable and uncoerced. And the 2024 PAGA reform makes Donohue-grade timekeeping the price of admission to its penalty caps: an employer seeking the 15 percent 'all reasonable steps' cap will be judged largely on whether its records could survive the presumption.

Donohue moved the center of gravity in meal period litigation from policy documents to punch data. A facially compliant policy now proves little; the case is won or lost in the time records, and every short or late punch without a premium or an attestation is a presumed violation the employer must explain. Plaintiffs' counsel builds entire PAGA demands from a punch-data export and the presumption — which is why the first substantive act in any meal period case, on either side, is the same records analysis.

The decision's interaction with the rest of the Laboratory is multiplicative. Each presumed violation carries a § 226.7 premium at the Ferra regular rate; unpaid premiums cascade into wage statement and waiting time exposure under Naranjo; and the aggregate becomes the PAGA penalty base the recoverability framework then disciplines. Donohue is the volume dial for all of it — the doctrine that determines how many violations the data will be deemed to show.

  1. Eliminate rounding of meal period punches wherever a system still does it — after Donohue there is no defensible version, and with Camp pending all rounding is a live risk.
  2. Audit punch data before the plaintiffs do: pull the unrounded records, flag every meal period under 30 minutes, after the fifth hour, or missing, and reconcile each flag against premiums paid and attestations captured.
  3. Deploy a contemporaneous attestation mechanism modeled on the dropdown Donohue itself blessed — an employee-facing prompt classifying every noncompliant punch as voluntary or not, with premium pay triggered automatically for everything else.
  4. Treat attestation records as litigation assets: they are the classwide rebuttal Donohue contemplates, and their absence leaves only employee-by-employee proof — decertification leverage in Allison, but no rebuttal at summary judgment or in PAGA.
  5. Rebut at scale with the tools the court preserved — representative testimony, surveys, and statistical analysis — rather than employee-by-employee declarations, and hold plaintiff surveys to Duran rigor, the standard the Allison survey failed.
  6. Strip rest period claims of punch-data inference — the presumption is records-based and rest periods need not be recorded — and force affirmative proof in place of statistical models built on meal-style assumptions.
  7. Use prospective written waivers for five-to-six-hour shifts under Bradsbery — drafted revocable and coercion-free — keeping signed, dated copies retrievable by employee for the full limitations period.
  8. Document the entire apparatus — unrounded timekeeping, attestations, premium triggers, waiver files — as the core of an 'all reasonable steps' showing under the 2024 reform's penalty caps.
Related Decisions
BrinkerFerraNaranjoTurrieta
This analysis is for informational purposes only and was last reviewed Q3 2026. Verify current status before relying on any authority.
Estrada v. Royalty Carpet Mills, Inc.Hohenshelt v. Superior Court