The objective of this study is to test the luminous efficiency functions V(lambda), V'(lambda), V(10)(lambda) and their linear combinations on the basis of a data set gained from a simulated mesopic night-time driving experiment. Another aim is to provide 'real-world' data for the 'X framework' or 'linear combination model', and to find out its limits in a practical situation. Human performance was measured by the reaction time method. Results show that the single parameter of the linear combination of photopic and scotopic luminous efficiency functions can be determined analytically with little variation for a given mesopic background luminance level and a given visual target colour, but the computation leads to considerable deviations comparing all three target colours (red, green and blue) used in the experiment. The conclusion for the given experimental conditions is that the single parameter of the linear combination model has an increasing deviation for lower background luminance levels.