Estimates of mean velocity and total driving time were studied for trips on an urban driving simulator. Routes traveled varied in two respects: the order of the segments, and the proportion of slow and fast segments. Regarding velocity, order effects were found: the first and last segments of a route influence mean velocity disproportionately. Beyond this, weighting by the proportion traveled at a given velocity is roughly correct, but there is a general leveling of differences as lower velocities are overestimated and higher ones underestimated. Total duration estimates were independent of order. Fast stretches are overestimated, slow ones are underestimated. These findings are explained by the attentional-allocation model of time perception, and a memory-based, retrospective estimation for mean velocity estimation.