I plan on compiling a short presentation for city council to see if we can do a better job of recording this information with emphasis that we are the worst in recording this data in the area.  The TxDOT database also included fatalities which tell some interesting stories as well.  Countywide about 7% of ped crashes included a fatality, for vehicle crashes it is less than 1%.

For nationwide data FARS would be the source, and I suspect it's similar but I haven't actually checked.

Alas, from what I've seen the only way to even come close to ensuring that a crash is properly investigated [edit: and documented/recorded] is for somebody to die ... and even that doesn't always work.

There might be more incidents but the data is not that great.  Looking at the TxDOT Crash Record Information System it appears 39% of pedestrian crashes and 24% of cyclist crashes do not have location data associated with them so it is difficult to pin point problem areas.  By comparison, less than 5% of vehicle crashes have no location data.  Does anyone know if this is typical for ped/cyclist crashes to have less location info?