In general, if conditions are equal, speed increases go hand in hand with more crashes and casualties; speed reductions with fewer crashes and casualties. Speed reductions or speed increases have the largest effect on the number of road deaths. Such changes have a slightly lesser effect on the number of serious road injuries, and lesser still on the number of slight injuries.
Approximately and on average, the following holds true: if the average speed on a road increases or decreases by 10%, the number of slight injury crashes rises or falls by 20%, the number of serious injury crashes by 30%, and the number of fatal crashes by 40%. This is a theoretical average based on kinetic laws [1]. As a formula, it looks like this:
Put into words: the ratio between the number of crashes before and after a change in speed equals the ratio between the average speed before and after that change to the power of x. Based on kinetic laws, for slight injury crashes this implies a power of 2, for serious injury crashes a power of 3 and for fatal crashes a power of 4.
Based on data of a large number of empirical studies into the effect of speed changes on crashes and also on a Power model, the exponents for different road types have been estimated [3]. This results in a ‘best estimate’ of the exponent. As can be seen in Table 1, the best exponent estimate for the number of road deaths is 4.6. To be 95% certain about the values the true effect lies between, we apply the formula twice, using the two bracketed exponents in the column headed 95% confidence interval – in this example, therefore, 4.0 and 5.2.
Crash/injury severity 
Rural roads 
Urban roads 

Best estimate 
Interval 95% confidence 
Best estimate 
Interval 95% confidence 

Road deaths 
4.6 
(4.0 – 5.2) 
3.0 
(0.5 – 6.5) 

Fatal crashes 
4.1 
(2.9  5.3) 
2.6 
(0.3  4.9) 

Serious injuries 
3.5 
(0.5  5.5) 
2.0 
(0.8  3.2) 

Serious injury crashes 
2.6 
(2.7  7.9) 
1.5 
(0.9  2.1) 

Sight injuries 
1.4 
(0.5  2.3) 
1.1 
(0.9  1.3) 

Slight injury crashes 
1.1 
(0.0  2.2) 
1.0 
(0.6  1.4) 
Table 1. Exponents in the formulas for the correlation between speed and crashes/casualties with different injury severity [3].
Reanalysis of the data [4] [13] shows that the exact correlation between speed and crash risk depends on the initial speed and can therefore be better described with an exponential model than with a Power model. Thus, a 10% reduction of average speed has a smaller effect when it concerns a reduction from 50 to 45 km/h than when it concerns a reduction from 100 to 90 km/h. In absolute terms, a speed reduction of e.g. 10 km/h will lead to a similar decrease of the number of crashes, independent of the initial speed [14].
There are no indications that the correlation between speed and road safety is less strong for newer cars, which are equipped with more systems to avoid crashes and which offer better protection to their occupants [15].