In this paper some implications of the pulse generator model of time perception have been tested. In the case of serial (re)production of an interval a lengthening effect occurs. Generally, this phenomenon is explained by assuming that the time-keeper is driven by the state of general physiological activation which decreases in the course of the task. An experiment was carried out in which EEG activity was recorded during the time estimation process. It appeared that a lengthening effect occurred, but there was no correlation between the reproduction times and the means of the EEG spectra, however. In a second experiment two series of time estimations were compared as a function of administering a placebo and methamphetamine. In the latter case the model predicts that no lengthening will occur since arousal (at least for a short period of time) is kept at a high and constant level. In fact, no differences between the regressions of the time estimation curves before and after drug administration have been found, so that the predictions of the model as regards the origin of lengthening could not be supported. It is suggested that the nature of the effect should be studied within the framework of the cognitive theory of time experience which offers various starting points in this respect.