The aim of this study was to investigate the relationship between the estimation of "time headway" and its real value in drivers of an automobile.　In the experiment the subjects who were driving an experimental vehicle, which was equipped with both GPS and laser radar, were required to follow another vehicle in front of them.　The "time headway" which was measured automatically by appropriate instruments was compared with the "time headway" of the subjective estimates from the subjects.　The estimated times of the subjects showed a high variability in all subjects; some subjects showed a significant variation of estimated time during the course of a day.　Furthermore, the estimates of the "time headway" were on a statistically level significantly inaccurate.　In other experiments which a "time headway" of 3 seconds or 4 seconds was required to be maintained by the subjects by counting aloud, thus, trying an adjustment to the subjectively perceived "time headway", it was observed that in such situations the actual "time headway" frequently was reduced by almost 1 second after counting.　In contrast, in the same situation, but with guidance provided by a clicking sound at regular intervals of 1 second, the perceived estimation of "time headway" improved to an acceptable degree of accuracy.