The identification of safe gaps between passing cars when crossing a street is a task most of us accomplish successfully on a daily basis. Objectively, how safe a specific gap is, is mainly dependent on how long it would take the approaching vehicle to arrive (time to arrival; TTA). Common sense might suggest that TTA is the basis for pedestrians' gap selection. However, it has been shown repeatedly that vehicle approach speed has a substantial influence on the size of chosen gaps. At higher speeds, pedestrians tend to accept smaller time gaps, i.e. they initiate riskier crossings. Some researchers have gone so far as to suggest that pedestrians rely more on physical distance of a vehicle in their crossing decisions than TTA. Yet, at the same time, there is evidence that TTA estimates themselves are influenced by object approach speed. It is suspected that pedestrians are more apt to base their decisions on systematically distorted TTA estimates, rather than physical distance. The goal of the two experiments described in this article was to explore the relationship between gap acceptance and TTA estimation. Participants were presented with video clips of approaching vehicles, and were either required to indicate a crossing decision, or to estimate TTA. Results show the typical effects of speed (smaller gaps at higher speed, lower TTA estimate at lower speed) and age (larger gaps for older participants). However, when using subjective time gap size (the TTA estimate) instead of objective time gap size to predict gap acceptance, the effect of speed either disappeared (Experiment I) or decreased substantially (Experiment II). The results indicate that systematic differences in TTA estimates can be a reasonable explanation for the effect of speed on gap acceptance.