How Likely Is a Second Cal/OSHA Citation?

A first Cal/OSHA citation raises a practical question: how likely is a second? Cal/OSHA publishes no answer. This page is CompliantCA’s, computed from the federal enforcement data: every California establishment first cited in 2020–2021, followed for three years.

The stakes are written in the penalty regulation. 8 CCR §334(d) defines a repeat violation as a substantially similar violation after an earlier cited violation occurring within the state, within five years of the first citation becoming final — and §336(g) multiplies a repeat’s penalty 2x, 4x, then 10x, with no abatement credit.

Free IIPP Audit Checklist. The records 8 CCR §3203 requires, one question each: can you produce it today? PDF, no account. Request the checklist →

The numbers

Of 8,144 California establishments whose first citation in the data landed in 2020–2021:

Three readings:

• The base rate is about one in eleven. It is not a follow-up artifact — removing follow-up inspections barely moves it — and a serious-class first citation nudges it up (10.2%).
• The 3.4% row is the legal-repeat shadow. §334(d) requires the second violation to be substantially similar; a later citation sharing a base standard with the first is the closest measurable proxy. Roughly a third of establishments cited again were cited again for the same kind of thing.
• The 18.9% row is mostly multi-location employers. 27% of the cohort carries a name that appears at more than one address — chains, franchisors, large firms. For them the statewide language of §334(d) matters: a citation at one site can set up a repeat classification at another.

Recidivism by sector

Construction’s low rate is partly an artifact of how we match. A factory can be revisited; a construction site stops existing when the job ends. Site-based matching therefore understates how often a contractor is cited twice — for construction, the same-name row above is the more honest measure of exposure, and §334(d)’s statewide rule means the moving jobsite is no shield.

Manufacturing’s 14.7% points the other way: fixed facilities with fixed hazards get return visits, and lockout/tagout — manufacturing’s second most-cited standard — carries the sector’s largest serious-class counts.

What this means for the second five years

Our three-year measurement is deliberately shorter than the law’s five-year repeat window, so every rate on this page understates the legal exposure period. The practical takeaway is the same one the penalty formula implies: the expensive citation is the second one. Fixing a cited condition durably — and keeping the records that show it stayed fixed — is what separates a $410 median first citation from a multiplied repeat.

Methodology — how we matched establishments

CompliantCA computed this analysis from the U.S. Department of Labor enforcement dataset (the OSHA Information System), which includes Cal/OSHA because California operates an OSHA-approved State Plan. Because the dataset identifies establishments by free-text name and address rather than a stable ID, the matching approach determines the numbers, so here it is in full:

• Establishment key: normalized name + five-digit zip. Names are uppercased, punctuation stripped, and trailing entity suffixes removed (Inc, LLC, Corp, Co, and similar), so “Acme Roofing, Inc.” and “ACME ROOFING INC” match. The zip keeps a chain’s locations separate — a second citation at a different location of the same company does not count in the same-location rows.
• Employer layer: normalized name only, reported separately, because §334(d)’s repeat rule is statewide. This layer over-groups genuinely distinct businesses that share a name and is presented as an upper-bound view of multi-location exposure.
• Cohort: establishments whose earliest citation in the data was issued in 2020–2021 (29,945 usable cited inspections; inspections with blank or “unknown” establishment names excluded; deleted citation records excluded throughout). The dataset begins in 2020, so “first citation” means first since 2020 — an establishment also cited in, say, 2018 would still enter the cohort.
• Outcome: any later inspection at the same key receiving at least one citation issued within three years (1,095 days) of the index citation. Every cohort member’s full three-year window closes before the data’s retrieval date (2026-08-21), so no rate is truncated. A sensitivity run excluding follow-up inspections is reported alongside the headline.
• Known limits: name changes, relocations, and successor entities break matches (understating recidivism); shared names over-group in the name-only layer (overstating it); construction’s transient sites understate its same-location row, as flagged above.

Cal/OSHA’s own statewide summaries are at DIR’s citation statistics pages. A statistic says what usually happened, not what will happen in any case — and nothing here is legal advice.