Let me get this straight. From the source dksuddeth used, Texas averages 1.25 "botched paramilitary police raids" per year out of over 1 million arrests. Bearing in mind that not all police actions even result in arrests, that means that speaking aggressively, .00000125% of arrest procedures are classified as "botched paramilitary operations".
Taking it a step farther, only a quarter of those result in the death of innocent persons. So, .00000031% of arrests end in the killing of innocent person.
Those figures, if accurate, reflect pretty well on police departments. I bet car manufactureres wish they could boast of safety margins like that. Hell, I wouldn't be surprised to find that more than .00000031% of dog owners are seriously wounded by their pets, or that .00000031% of surgeries result in fatal mistakes. After all, that's basically one in 4 million.
Coincidentally, 1:4,000,000 are the odds that you'll match all 5 numbers in the Texas Mega Millions drawing.
EDIT:
I did a quick check of the medical mistake numbers.
Using figures from
http://www.wrongdiagnosis.com/mistakes/common.htm and
http://www.dshs.state.tx.us/CHS/VSTA...st/ageadj.shtm, you can find that in 1999, 2700 people may have died due to some sort of medical mistake. This means that for that year, assuming the trends were relatively stable, you were 440 times more likely to be killed by your doctor than by the police.