Pedestrian gap acceptance has not been explored to the same degree as vehicle gap acceptance. Although the two are similar in concept, there are a variety of pedestrian characteristics and caveats in the interaction between the pedestrian and vehicle modes that require separate pedestrian gap acceptance models. The objective of this research was to analyze empirical observations collected at 27 sites in Alabama, Florida, and North Carolina and to develop pedestrian gap acceptance models at midblock crossings. Goodness-of-fit tests showed that a probit-based, single-lane gap acceptance model, drawn from a noncontrolled pedestrian crossing data set, yielded the best results (max-rescaled R2 5 .69). This model involved only two parameters: The size of the gap length in seconds and a binary variable that distinguished between gaps and lag events (first-arriving vehicle, without a prior lead vehicle, to open the gap). An increase in gap length was associated with an increased probability that a pedestrian would cross, whereas a lag event had a negative coefficient, which meant that a pedestrian was less likely to accept a lag than a gap, given the same length in seconds. Other variables that described crosswalk characteristics as well as pedestrian and driver behavior did not emerge as significant factors in the model. The research reported in this paper offers a new, robust pedestrian gap acceptance model that can be used in traffic operational analysis. The model can be incorporated into microsimulation packages and thus improve the accuracy of pedestrian behavior modeling at midblock crossings in the future.