Assetto Corsa EVO safety rating: proximity that pays

For years, protecting your safety rating meant one thing: giving way. Lifting off as you approached a pack, letting others through, avoiding risk. The Assetto Corsa EVO safety rating turns that logic on its head — it measures the time you spend fighting in close proximity without making contact. This article explains what the system actually looks at, how it differs from what you know of iRacing, and what that means in practice for the way you race this autumn.

What the Assetto Corsa EVO safety rating measures

Kunos summed up the philosophy behind their system, called EVO SR, in one sentence worth quoting directly: "Great racing is measured by how you race others, not how well you avoid them."

In practice, the system measures the time spent racing close to other cars without contact. It is not an incident counter with the sign reversed: it is a measure of presence. Staying in the fight, holding a wheel on the inside, exiting a corner side by side — that is what feeds the rating. A clean lap, alone, thirty seconds behind the nearest rival, proves nothing to the system and earns nothing from it.

This shift in perspective is worth understanding before you take a start. The question is no longer "did I stay out of trouble?" but "did I spend time where races are decided, and did I hold my own there?"

Why EVO SR takes the opposite stance to iRacing

If you come from iRacing, your instincts have been shaped by years of inverse logic. There, the safety rating drops on contact, and it often drops for both cars involved — the one that pushed and the one that was pushed into. The rational calculation for any driver looking to move up a category is straightforward: the less contact, the better. Many end up racing in clear air for purely arithmetical reasons.

In Assetto Corsa EVO, that arithmetic no longer holds. Keeping your distance does not raise your rating — it simply lets it sleep. The driver who spends an entire race running in empty space and the one who fought fifteen clean laps in a battle do not leave with the same credit, and it is the second whom the system recognises.

There is an invitation here that deserves to be taken seriously: proximity ceases to be a risk to avoid and becomes a skill to demonstrate. For many drivers, that means relearning the fundamentals of pack positioning from the ground up.

The system reads the point of impact, not merely the collision

The objection arises immediately: if you reward proximity, you also reward the contact that comes with it. Kunos addressed this concern on two fronts.

First, attribution of responsibility. The system analyses points of impact to determine who caused the contact — the stated intention being that "aggressors pay and innocent victims do not." That is a clear departure from a model that penalises both cars in an incident indiscriminately.

Second, severity. Not all contact is equal. A door-to-door brush through a fast corner has only a marginal effect on the rating, whereas a serious collision carries significant weight. The system therefore draws a distinction between racing contact, which belongs to the sport, and a driving offence, which does not.

A word of honesty on this point: deciding algorithmically who is responsible for an incident remains genuinely difficult. Experienced stewards do not always agree with one another on the same moment. We will judge this system on the evidence of racing, not on its announcement.

What our telemetry sees — and what it will never see

Here is the concrete example, and it comes from our own data. The trace below is taken from JOBARD Race Engineer, from a race run by the team at Road America in the Chevrolet Corvette Z06 GT3.R. It comes from iRacing, not from Assetto Corsa EVO — we say so plainly, because it is precisely what it demonstrates that matters here.

Nineteen laps covered. Fifteen valid, four invalidated. Two stints separated by a pit stop. And a very clear progression across the first stint: 2:07.171 on lap two, then 2:05.720, 2:05.701, 2:05.532, and finally 2:05.156 on lap ten — after which no further personal best across the remaining nine laps. The driver found the limit within eight laps, then held it. That is exactly what a controlled stint looks like.

Now, the interesting part: nothing in those figures says how close to any rival those laps were completed. Not a single data point. A racing telemetry system reads your car — speed, braking, steering angle, delta to the reference lap. It is blind to everything else.

In other words, what EVO SR measures is precisely what no solo telemetry tool can see. Your four invalidated laps show clearly on a trace; your fifteen laps spent defending the inside line do not. This is good reason not to conflate the two disciplines: telemetry makes you faster, the safety rating attests to how you race in company. You need both.

What is not public, and what we will not claim

Let us be clear about the limits of what is known. To date, neither the numerical scale of EVO SR, nor its bands, nor the access thresholds for ranked servers have been documented publicly by Kunos with the precision a driver would want. You will find correspondence tables and detailed breakdowns elsewhere — we will not publish any until they are properly sourced.

What is established: the philosophy, the measurement of proximity without contact, the attribution of responsibility by point of impact, and the weighting by severity. That is already enough to change the way you approach a start.

Key points - EVO SR rewards the time spent wheel to wheel without contact — staying clear earns nothing. - The system reads the point of impact and the severity: a racing brush does not cost what a driving offence costs. - No solo telemetry tool measures what EVO SR records. The two are developed separately.

Eight rounds to learn it in competition

You do not build your safety rating by driving alone: by design, it is decided in the company of others. That is the purpose of the JOBARD Racing Academy Series — eight free rounds on Assetto Corsa EVO, from September to December 2026, broadcast on Twitch. A regular framework, rivals of the same level, and a team's perspective on your progress round by round. Exactly the environment where EVO SR is built.

Discover the JOBARD Racing Academy Series →

Two further reads complement this one: our guide to your first month on Assetto Corsa EVO, which lays the groundwork before your first race, and the five telemetry graphs that drive improvement, for the lap time side of things. If your challenge is holding the apex in a pack, start instead with braking later is not enough.

In brief

Does the Assetto Corsa EVO safety rating work like iRacing's? No, and that is the central point. iRacing counts incidents; EVO SR measures the time spent racing close to others without contact. In EVO, avoiding the fight does not raise your rating.

How do you improve your EVO SR? By taking part in battles and holding them cleanly. The system values presence in the pack, weights the severity of contact, and looks to assign responsibility rather than penalise both cars.

Does any contact automatically lower the rating? No. A door-to-door brush has a marginal effect; a serious collision carries considerably more weight. And if the point-of-impact analysis identifies you as the victim, you are not expected to pay for another driver's mistake.

Sources: Assetto Corsa EVO — EVO SR, official Kunos Simulazioni announcement, 17 June 2026 · Traxion and BoxThisLap coverage, May 2026 · JOBARD Race Engineer data, Road America session, 25 July 2026 (Chevrolet Corvette Z06 GT3.R, iRacing).