The difficulties of deciding whether subjective time grows as a linear or nonlinear function of real time are discussed, and two experiments are presented to address this question. In Experiment 1, people received a 10-s standard duration and then had to judge what proportion other durations (ranging from 1 to 10 s) were of the standard. Counting was prevented by a concurrent task. The relation between judged and actual proportions was linear. In Experiment 2, people were required to average together three tone durations (mean duration 600 ms) and to judge whether subsequently presented comparisons were or were not the average. The spacing of the tone durations had no effect on judgements, suggesting a linear underlying time scale.