How constant is the Weber fraction (WF) for brief time intervals? This question was assessed in three experiments with two base durations (BDs), 0.2 and 1 sec, and with different ways of estimating the WF. In Experiment 1, the psychometric functions were drawn on the basis of 4, 8, or 12 comparison intervals with the shortest to longest duration ranges being kept constant. The results revealed no effect of the number of intervals, but the WF (threshold/BD) was significantly lower at 0.2 sec. In Experiment 2, the comparison intervals were distributed over three duration ranges. There was no range effect, and the WF was generally lower at 0.2 sec than at 1 sec. In Experiment 3, one condition allowed a comparison of the BD with the same range between the shortest and longest comparison intervals. Once again, the WF was lower at 0.2 sec than at 1 sec. Overall, the results reveal (1) that increasing the number of comparison intervals or the duration range does not seem to affect the value of the WF and (2) that the WF is lower at 0.2 sec than at 1 sec, which is inconsistent with the scalar property of some timing models.