One serious crash every eight days

Waymo's driverless fleet has crossed 170 million rider-only miles — roughly 200 human lifetimes of driving, accumulating at over four million miles a week. Across that span the company's latest safety analysis counts 35 fewer serious-or-fatal-injury crashes than it estimates human drivers would have had on the same streets: a 92% reduction, up from the 85% it reported at 56.7 million miles. Scaled to the current pace, Waymo frames it as one serious crash prevented every eight days.

The whole safety case rests on 35 serious crashes that, by Waymo's own count, would otherwise have happened — a small number carrying a 92% claim.

The number that matters is the one Waymo built itself. There is no public ledger of the crashes that didn't happen, so the 92% rests entirely on a counterfactual: state police records and traffic volumes for the exact counties Waymo operates in, surface streets only, marked up 32% to cover crashes humans never report. The earlier 56.7-million-mile version of this analysis was peer-reviewed; the 170-million update is not yet. And the whole civilizational headline rests on a small absolute count — 35 serious crashes that, by Waymo's reckoning, would have been more.

What's genuinely new is that the figure is firming as the miles pile up rather than reverting toward the human baseline — the pattern you'd want before trusting a safety claim. The honest reading sits between the press release and the skeptics: this is the strongest evidence yet that removing the driver lowers serious-crash risk at real scale, inside a geofence, on Waymo's own scorecard. The open question is whether the edge survives contact with freeways, weather, and an auditor who isn't Waymo.