Black, grey, silver, red and blue fail to stand out against the background of the road, scenery and other traffic.

The team, from Monash University in Australia scrutinise police data on 850,000 accidents for information on the car, they time of day and the type of prang.

Commercial vehicles, including taxis and white vans, were excluded from the mix.

After trying to take into account the possibility that drivers who take risks might be drawn more to some colours than others, they found black cars to be most accident prone.

During daylight hours, they were up to 12 per cent more likely be in crashes than white vehicles, while at dawn and dusk, the figure rose to 47 per cent.

Grey and silver cars were the next most risky, followed by red and blue, the journal Safety Science reports.

Interestingly, the study hinted that orange cars may be even safer than white ones - although the scientists could not be confident about this.

They said that although their findings seem to suggest that steering customers away from black cars would make the roads safer, this might not necessarily be the case.

This is because having lots more white cars on the road would mean there was less contrast between vehicles and so make them harder to spot.

Researcher Dr Stuart Newstead said: 'Whilst campaigns to modify vehicle colour choice could alter the crash risk for the fleet, colour is a much less influential crash risk modifier than behavioural traits such as drink-driving, and speeding.

'It may be possible that simple solutions such as the use of daytime running lights or headlights could effectively negate the elevated risks of higher risk vehicle colours.'

Read more: http://www.dailymail...l#ixzz0rZuGBMuV

This may be a case of the garbage in, garbage out principle very common in computer modelling also showing up in statistical analysis. Put enough data together and put it through enough filters and you'll get correlations whether they exist or not. The question is how do they remove all the other factors from the analysis without adding even more bias? I've driven green, black, wine, grey and blue cars and not noticed any change in the number of near misses/accidents. The only vehicle I've driven where I have seen a difference is a white van, but given it's 7.5 meters long and about the biggest thing that can be driven on a car licence, that's to be expected...