Models using contrast as the measure of discernibility or detection of roadway obstacle on non-lighted roadways have not recognized the lighting effects caused by the lateral spacing of headlamps on the vehicle. When the only significant sources of roadway light are a vehicle's headlamps, shadows on the roadway behind and to the left and right of the obstacle are viewed simultaneously with the obstacle itself. It can be shown that in certain obstacle-sighting situations, where there is little or no significant background light from sources other than the headlamps themselves, sufficient contrast is provided by the shadow outlines. Shadows created by the lateral spacing of the headlamp light sources and the driver's eyepoint, framing one or both sides of the obstacle, serve to outline the illuminated obstacle on the roadway in front of a vehicle. In the context of detecting a roadway obstacle, the contrast requirements are satisfied (or obviated) and threshold luminance becomes a relevant measure of obstacle discernibility. Comparison of the results of field experiments with the predictions of both contrast predictive and luminance predictive models shows that, within the limits of the situations tested, the luminance model gives a better measure of detection distance and obstacle luminance at detection.