These are counts of people, not rates of risk. June has more casualties than January partly because far more people ride in June. Nothing on this page says a sunny afternoon is more dangerous per mile than a wet night, and DfT does not publish exposure by month, day or hour, so no such rate can be worked out from this data.
Deaths per 100 casualties is a different measure and it does not have that problem. It only compares people who were already in a reported collision, so it answers a narrower and more answerable question: once something has gone wrong, how likely is it to be fatal.
Police data undercounts injured riders, and undercounts injured cyclists more, because a cyclist has no obligation to report a collision. Treat every count here as a floor.
Severity is as reported by the police and is not severity-adjusted. DfT publishes an adjusted figure to account for changes in police reporting systems since 2016, which is why its 2025 total of 5,710 seriously injured motorcyclists differs from the 5,616 counted here. Both reconcile to the same 386 deaths. We use the unadjusted count so that this page and our most dangerous roads page count the same thing.
Weather, light and road surface are recorded once for the collision, not for each casualty.