The safety triangle is a model that stacks minor incidents, near misses, and unsafe acts beneath every serious injury to argue that controlling the wide base of small events reduces the rare catastrophic one at the top. Two versions dominate the conversation: Herbert Heinrich's 300-29-1 ratio from 1931 and Frank Bird's 600-30-10-1 ratio from 1969. They look similar, but they were built from different data, count different things, and imply different priorities. Knowing which one to trust (and when to trust neither blindly) changes how you invest in reporting and prevention.
Heinrich studied roughly 75,000 insurance and industrial accident records and concluded that for every major injury there were about 29 minor injuries and 300 no-injury accidents. The famous shorthand is 300 : 29 : 1. His deeper claim, and the more controversial one, was that roughly 88 percent of accidents traced to unsafe acts by workers rather than unsafe conditions.
Frank Bird analyzed about 1.7 million incident reports from nearly 300 companies and produced a four-tier pyramid: 600 : 30 : 10 : 1. Read from the top down, that is 1 serious or disabling injury, 10 minor injuries, 30 property-damage accidents, and 600 near misses (incidents with no visible injury or damage). Bird's addition of a distinct property-damage tier and a large near-miss base was the real innovation, because it put uninsured and hidden costs on the map.
The two triangles are not describing the same categories, so lining them up side by side is misleading. Heinrich's "300" are no-injury accidents. Bird's "600" are near misses, and he also carved out 30 property-damage events that Heinrich never separated. When someone says "the ratio proves near misses matter," they are usually quoting Bird, not Heinrich, even if they cite the wrong name.
Both are population averages across many industries and eras. Neither was ever meant to predict your specific plant, and modern researchers have shown the ratios vary enormously by sector, hazard type, and how diligently near misses get logged.
Suppose your site recorded 2 serious injuries last year. If you took Bird's 600-30-10-1 literally, you would expect the base to look like this:
Now compare that to what you actually captured. If your log shows 2 serious injuries, 18 minor injuries, 12 property-damage events, and only 40 near misses, the mismatch is the signal. Injuries roughly track the model, but you are capturing 40 near misses where the ratio predicts around 1,200. That does not mean your floor is 30 times safer than average. It almost certainly means near misses are going unreported. The triangle here is not a prediction to hit; it is a gap detector for your reporting culture. Chasing the missing 1,160 reports (through easier logging and blame-free follow-up) is usually a better bet than declaring victory.
Treating either triangle as gospel is where teams get burned. The core critique, backed by decades of later research, is that reducing minor incidents does not reliably reduce fatalities, because serious and fatal events often have different causal pathways than slips and small cuts. A workplace can drive its minor-injury rate to near zero while its exposure to a low-frequency, high-consequence hazard (a confined-space release, a fall from height, a machine guarding failure) stays fully intact.
Heinrich's 88-percent "unsafe acts" figure has aged worst of all. It nudges organizations toward blaming individuals instead of fixing systems, conditions, and design. Modern safety thinking, and structured methods like 8D problem solving and A3 problem solving, push you toward root causes in the system rather than a tally of worker mistakes. For hazard identification, dedicated techniques such as HAZOP and FMEA map failure modes far more usefully than a ratio ever could.
Use Bird's four-tier model as the practical default, because separating property damage and near misses reflects how factories actually generate data, and it justifies investment in near-miss capture. Use Heinrich's history as a caution, not a target: keep the idea that a broad base of small signals precedes big events, but drop the fixed ratio and the "unsafe acts" framing. In short, trust the direction both models point (small signals matter) and distrust the precise multipliers and any claim that minor-incident reduction guarantees fewer fatalities.
The productive move is to treat the triangle as a prompt to build a reporting system, then prioritize high-consequence hazards separately. Pair leading indicators (near-miss rate, closure time on unsafe conditions) with a hard focus on the handful of ways someone could actually be killed or maimed on your site.
The safety triangle only works if the base of the pyramid is real data, and most sites lose near misses because logging is friction. Fabrico is the real-time data foundation that makes capture practical. Its field-ready CMMS lets operators and technicians raise work orders, log unsafe conditions, track assets and spare parts, and run proactive maintenance schedules so hazards get fixed before they climb the triangle. Because Fabrico also delivers real-time OEE and production monitoring (including computer vision on machines with no PLC), the same stops and anomalies that hurt throughput often surface the equipment risks that safety reporting misses. You can then apply Pareto analysis to the logged events to concentrate effort where it counts. Fabrico is EU-built with EU data residency, so all of that reporting stays on infrastructure you can point to. See the CMMS solution overview for how the reporting and work-order flow works in practice.
As a rough qualitative idea (many small signals precede big events) it holds up. As a precise predictive ratio it does not. Peer-reviewed work has repeatedly shown the multipliers vary by industry and that lowering minor-injury counts does not reliably lower serious or fatal injuries, because those often have separate causes.
Prefer Bird's 600-30-10-1 for day-to-day practice because it separates property damage and near misses, which matches how factories generate data and justifies near-miss reporting. Use Heinrich mainly as historical context, and avoid his "unsafe acts" blame framing entirely.
Many workplace incidents trace back to equipment condition, so a strong maintenance record feeds the same base. Tracking MTBF and MTTR and moving toward condition-based maintenance shrinks the population of failures that would otherwise become near misses or injuries.
Want to turn near misses, unsafe conditions, and equipment faults into one live record your team actually keeps up to date? Book a Fabrico demo and see how real-time OEE and a field-ready CMMS give you the reliable base your safety data depends on.
Login
Blog
