A plain-English breakdown of how the premium is built, what it averages per $100 of payroll and per employee in 2026, and the industry and state factors that move the number — with a worked example and a free calculator.
"How much is workers' comp?" has no single answer, because the price is custom-built from your payroll, your industry's injury risk, your state's rules and your own claims history. A clerical-only consultancy might pay well under $0.20 per $100 of payroll; a roofing crew can pay $20 or more per $100 — a hundred-fold spread for the same coverage. This guide shows exactly how that number is assembled so you can estimate yours and see where you can lower it.
Almost every US workers' compensation premium follows the same core formula, standardized by the National Council on Compensation Insurance (NCCI) in the 38 states it serves and mirrored closely by independent rating bureaus in states like California (WCIRB), New York (NYCIRB) and Pennsylvania:
| Component | What it is |
|---|---|
| Payroll ÷ 100 | Rates are quoted per $100 of gross payroll, so payroll is the exposure base. |
| × Class-code rate | Each NCCI/bureau class code (e.g. 8810 clerical, 5645 carpentry) has its own rate reflecting that work's injury cost. |
| = Manual premium | The starting "rack rate" before any adjustment. |
| × Experience modifier (EMR) | Your claims history vs. peers. 1.0 is average; below 1.0 is a credit, above 1.0 a surcharge. |
| × Schedule / premium credits | Underwriter discretion, safety-program credits, premium discounts for size. |
| + Expense constant | A small flat fee (often $150–$250) added to every policy to cover issuance costs. |
In short: (payroll ÷ 100) × class rate × EMR, then credits/debits and the expense constant. The two levers you do not control are the class rate (set by the bureau and state regulator) and your state. The lever you do control is the EMR — and over time, your class assignment by managing how work is categorized.
Honest answer: averages are almost useless for any single business, but they set expectations. National data points for 2026 cluster like this:
Treat any single "average" with suspicion. A figure like "$1,350 per employee per year" is the mean of a clerk costing $200 and a steelworker costing $4,000 — the average describes neither. The spread is driven almost entirely by class code and state, so use the formula above with your real class code instead of anchoring on a national mean.
Industry is the single biggest driver because the class-code rate is built from that occupation's historical injury and claim costs. Indicative rate ranges per $100 of payroll (they vary widely by state and year):
| Work type | Typical rate / $100 payroll | Why |
|---|---|---|
| Clerical / office (8810) | ~$0.10–$0.30 | Low-energy, seated work; few serious claims. |
| Retail / restaurant | ~$0.50–$2.00 | Slips, burns, cuts, lifting. |
| Manufacturing | ~$1.50–$6.00 | Machinery, ergonomics, materials handling. |
| Trucking / delivery | ~$4.00–$10.00 | Road risk, heavy loads, long shifts. |
| Construction / carpentry | ~$5.00–$15.00 | Falls, struck-by, energized work. |
| Roofing (5551) | ~$15.00–$40.00+ | Highest fall-from-height exposure of any common code. |
This is why payroll for a single business is split by class code: an office employee on a construction firm's books is rated at clerical, not at the framing rate — provided the payroll records cleanly separate the duties. Misclassifying high-hazard payroll as clerical is a common audit finding and triggers back-premium and penalties.
The same roofer can pay double in one state versus another. State rules on benefit levels, medical fee schedules, litigation environment and claim duration all feed the class rates that state's regulator approves. California and a handful of others have historically run among the highest rates; states like Indiana, Arkansas and several in the South tend toward the lower end. Rates are revised annually, so always check the current bureau filing.
Monopolistic states are a special case. In four states — North Dakota, Ohio, Washington and Wyoming — you cannot buy workers' comp from a private insurer at all. You must purchase it from the state-run fund (and in Washington, premiums are partly calculated on hours worked rather than payroll, and part of the cost can be deducted from employee pay). Because their experience-rating systems differ, an EMR or quote logic from an NCCI state does not transfer to these funds. A few other states run a competitive state fund alongside private carriers, which can be an option if private quotes come back high.
Within the formula, here is where the money actually moves:
| Raises cost | Lowers cost |
|---|---|
| EMR above 1.0 from frequent or severe claims | EMR below 1.0 from a clean, well-managed loss record |
| High-hazard class codes and growing payroll | Accurate payroll split so low-risk work is rated low |
| Operating in a high-cost state | Documented safety program earning schedule credits |
| Misclassification corrected at audit (back-premium) | Return-to-work program that cuts claim duration and cost |
| Owner/officer payroll included unnecessarily | Excluding eligible owners/officers where state law allows |
The biggest controllable lever is the experience modifier. Driving your EMR from 1.15 down to 0.90 cuts premium by more than 20% on the same payroll — which is why injury prevention, prompt claim handling and a return-to-work program are direct cost-control tools, not just compliance.
Every policy carries a minimum premium — often a few hundred dollars — so even a tiny payroll won't produce a $30 bill. For sole proprietors and single-member businesses with no employees, insurers sell a ghost policy: a minimum-premium policy where the owner is excluded from coverage. It exists purely so the owner can hand a certificate of insurance to general contractors and clients who require one. Important caveat: a ghost policy does not cover the owner if they are hurt — it only proves coverage exists for any employees they might add.
The premium you pay at the start of the policy is an estimate, based on the payroll you project for the coming year. At expiry the insurer runs a premium audit: it reconciles your actual payroll, by class code, against that estimate. If you paid out more wages than projected — or had high-hazard work that was booked at a clerical rate — you receive an additional premium bill. If you ran lower, you get a refund. This is why the single most expensive mistake small employers make is under-reporting or misclassifying payroll: it does not save money, it just defers the cost to audit and adds interest and penalties. Keep clean payroll records that separate duties by class code, document any subcontractors (uninsured subs are usually charged back to you as if they were your employees), and respond to the audit promptly. A cooperative, well-documented audit is one of the cheapest ways to keep your effective rate honest.
Two related details that surprise new buyers: overtime is generally reported at base (straight-time) wages, not the time-and-a-half rate, so paying overtime correctly avoids inflating your exposure base; and payroll caps apply to high earners and owners in many states, limiting how much of an executive's wages count toward premium. Both can meaningfully lower the audited figure if applied correctly.
A small framing contractor with a foreman, an office admin and three carpenters in an NCCI state:
| Payroll segment | Annual payroll | Rate / $100 | Manual premium |
|---|---|---|---|
| Carpentry crew (4 incl. foreman) | $240,000 | $9.00 | $21,600 |
| Clerical admin (1) | $45,000 | $0.25 | $113 |
| Manual premium subtotal | $21,713 |
Now apply the experience modifier and expense constant. With a clean record giving an EMR of 0.90:
Had the same firm carried an EMR of 1.20 from two lost-time claims, the math becomes $21,713 × 1.20 = $26,056 — roughly $6,300 more per year for identical work. That gap is the dollar value of a functioning safety program. Notice too how little the office admin adds ($113) versus the crew ($21,600): the carpentry class rate, not headcount, dominates the bill.
Don't guess from a national average. These free, no-signup tools run entirely in your browser:
What is the average cost of workers' comp per employee?
Across all US industries it lands near roughly $40–$95 per employee per month (about $500–$1,150 a year) — but that blends a clerk under $20/month with a roofer costing several hundred. Your real number depends on payroll, class code, state and EMR.
How is a workers' comp premium calculated?
(Annual payroll ÷ 100) × class-code rate × experience modifier (EMR), then schedule credits/debits and a flat expense constant. Each job class has its own rate per $100 of payroll.
What raises my workers' comp cost the most?
A high-hazard class code, an EMR above 1.0 from claims, a high-cost state, misclassified payroll and growing payroll. Cutting injuries to push your EMR below 1.0 is the biggest lever you control.
What is a ghost policy?
A minimum-premium policy for a business with no employees but the owner (who is excluded). It lets the owner show a certificate of insurance to clients and GCs, but it does not cover the owner if injured.
Which states won't let me buy from a private insurer?
Four monopolistic states require coverage from a state fund: North Dakota, Ohio, Washington and Wyoming. Everywhere else you can buy from private carriers.
This guide is general information, not insurance, legal or tax advice, and AEGIS - AMA is independent of any insurer. Rates, formulas and state rules change every year — confirm current figures with NCCI, your state bureau or fund, and a licensed broker before relying on them.