In 2002/3 I carried out a review of daytime running lights (DRL) for the NRMA (a motoring organisation in New South Wales). At the start of the project  I was skeptical about their effectiveness in Australa. However, I soon located a 1991 CIE report that made a strong case for well-designed DRLs. CIE is the international authority on lighting standards. I applied the CIE recommendations to some work I carried out on flashing lights on school buses with Dr Alec Fisher and realised that low beam headlights (maximum luminous intensity 437cd in direction of oncoming motorists) could only be expected to be marginally effective during all but very dull days. This helped explain the so-called "latitude effect" where studies seem to show that DRLs are more effective in high-latitude locations like Scandinavia. Bright DRLs (up to 1200cd, as recommended by CIE) would be effective under all daylight conditions and so were better suited to Australia.

The latest designs of dedicated DRL are very promising - they send light is the optimum direction and consume very little power (just 12W in the case of a Hella kit). They dispel most of the criticisms that have been leveled against DRLs.

The instant you see headlights at night you know that a vehicle is approaching. DRLs work in the same way during the day - separating an approaching vehicle from visual clutter.

This page contains links to DRL studies and some comments on those studies.  A key point is that most effectiveness studies are based on a large proportion of vehicles with low beam headlights but photometric analysis suggests that these are only marginally effective on all but very dull days. A 2003 GM/SAE paper confirms a large difference in effectiveness for the various implementations of DRL This is explored further in my paper for the 2005 Enhanced Safety of Vehicles Conference.
                   [Apologies that some links no longer work]
Comments on ATSB report:
My analysis of signal range suggests that DRL can be designed to be effective under "bright" daylight conditions. I consider this worthwhile because many accidents still  occur in bright daylight where the approaching vehicle was not seen (although this has not been quantified). Days with a light cloud cover in Australia are typical conditions where constrasts are poor (no shadows) and bright DRLs could be expected to be quite effective. I  did suggest that that a light sensor could be used to increase DRL intensity in bright conditions but did not try to analyse the effects:

Page 26 "Control with light sensors - Many Canadian vehicles have ambient lighting sensors that switch from DRLs to headlights
automatically when light levels fall (ambient light on a horizontal surface less than 1000 lux according to SAE J2087). This overcomes the criticism that DRLs could be inadvertently used at night and cause undue glare. The cost of adding light sensor control is relatively small,
compared with the cost of a complete DRL installation. It appears feasible to adapt this technology to increase DRL light intensity (beyond the 1200cd recommended by CIE) under very bright conditions when glare is not an issue."

I do not agree that extinguishing DRLs on bright days (to save energy) is worthwhile.

References and updates

The best DRL solution is low-powered (LED?) dedicated DRLs with a light sensor to switch to headlights when ambient light levels drop.
Illustrations of DRLs in "action"
Matrix snapshot

Maxtrix snapshot

Note that lack of DRLs on the motorcycle. Many cars in the USA use bright front turn signals as a DRL. I have suggested that motorcycles should do the same. This would use no more space than current turn signals.

The animated GIF (click on image below) illustrates the difference between ordinary turn signals and bright DRL turn signals on a motorcycle. The images have been digitally enhanced.
Click to animate

400K Quicktime movie