How fault is determined — the words that decide it, and who says them

“It was obviously his fault. Why is the insurer even asking?”

Because fault, in a claim, is not a judgment about who drove worse. It is an allocation of legal responsibility, made from a small vocabulary of defined terms, and the person making it first is almost never a judge. It is an adjuster, reading a file, applying the rules of a system you may not know you are in. Most claimants argue about the facts when the disagreement is actually about the words. So take the words first.

Negligence is the base term. It means failing to take the care a reasonably careful driver would have taken in the circumstances, and causing harm by that failure. It is not recklessness, it is not intent, and it does not require anyone to have been driving badly in a memorable way. A momentary lapse is negligence if a careful driver would not have made it.

Duty, breach, causation, damage are the four things that have to be present for negligence to produce liability. Every driver owes other road users a duty of care; a breach is the failure to meet it; causation links the breach to what happened; damage is the loss that follows. The chain matters, because a claim can fail on causation even where the breach is obvious — if the other driver was speeding, but the collision would have happened anyway, the speeding may not be what caused it.

Contributory negligence is your own share of the failure. In a small number of systems it retains its historical force: any contribution by the claimant, however small, bars recovery entirely. In most systems it has been replaced by the next term.

Comparative negligence apportions fault between the parties as percentages and reduces each party’s recovery by their own share. It comes in two shapes. In the pure form, you recover your share of the loss no matter how large your fault; a claimant found mostly responsible still recovers something. In the modified form, recovery is reduced in the same way but stops entirely once your share crosses a bar that the jurisdiction sets. The difference between those two shapes is, in practice, the difference between a partial payment and nothing, and it is decided by where the crash happened rather than by anything you did. Which rule your jurisdiction uses is in the data below.

No-fault describes a different arrangement altogether: your own insurer pays defined losses — typically medical costs and lost income — regardless of who caused the crash, and your right to sue the other driver is restricted until the injury passes a threshold the law sets. No-fault does not mean nobody was at fault. It means fault has been removed from the first stage of the process, not from all of it; property damage and serious injury generally return to the ordinary rules.

Apportionment is the act of assigning the percentages. It is the step where a claim quietly loses money, because a shift from “not at fault” to “twenty per cent at fault” reduces everything downstream without anyone denying the claim.

Who actually decides

Now the subject, in those terms. Almost every fault determination is made by insurance adjusters, not by courts. Your insurer forms a view and the other insurer forms a view; where they agree, that is the outcome, and nothing is adjudicated. Where they disagree and the money is small, they may settle at a split neither believes precisely, because litigating costs more than the gap. Where they disagree and the money is large, it goes to a lawyer, and eventually — rarely — to a judge who applies the same vocabulary with more rigour and a better view of the evidence.

This matters for a practical reason. If you are arguing about fault, you are arguing with a person who has a file, a caseload and a set of internal rules, and who is persuaded by documents rather than by indignation. The things that move an apportionment are the police report and any citation issued, the physical damage and where it sits on each vehicle, independent witnesses, video, the vehicle’s own recorded data, and the consistency of each driver’s account with all of the above. What does not move it is certainty, volume, or the fact that you know you were driving carefully.

Presumptions, and why they are not rules

Certain configurations carry strong presumptions: the following driver in a rear-end collision, the driver turning across oncoming traffic, the driver emerging from a minor road. These are not automatic determinations. They are shortcuts that reflect where the duty usually sits, and each can be rebutted by evidence that the duty sat elsewhere — brake lights that did not work, a vehicle that reversed, a driver who stopped in a live lane for no reason. Treating a presumption as a rule is the commonest way claimants are surprised late.

What is contested

Two disagreements are live and worth knowing about.

The first is about recorded vehicle data. Insurers and reconstruction specialists argue that event-data recorders and telematics give an objective account of speed, braking and steering in the seconds before impact, and that this is better than two contradictory memories. Privacy advocates and defence lawyers argue that the data is collected under consents drivers do not read, that it is interpreted selectively, and that a few seconds of vehicle telemetry says nothing about what the other car did. Both are right about something: the data is more reliable than memory, and it answers a narrower question than the one being asked.

The second is about the comparative-negligence bar itself. Its defenders say a driver substantially responsible for their own injuries should not be subsidised by someone less responsible. Its critics say a bar of that kind turns a percentage judgment made by a human being into a cliff, so that a point on either side of it is the difference between compensation and none. This one has no technical answer; it is a policy choice, made differently in different places, and it is the reason the same crash pays out differently across a state line.

What we do not know

We cannot tell you what an adjuster will conclude about your crash, and anyone who tells you a percentage before seeing the file is guessing. We do not know how heavily any particular insurer weights recorded vehicle data against witness accounts, because those internal standards are not published. And where a liability model is used to support the determination, we generally cannot tell what it was trained on or how much weight the adjuster gave it — which is, at the moment, the most consequential unknown in this part of the claim.

What you can do about any of it is narrow but real: establish the facts early, in documents, while they are still recoverable. Fault is decided from the record. The record is built in the first week, mostly by other people.