This paper aims to explore pedestrian temporal and spatial gap acceptance at uncontrolled mid-block street crossings, where vehicles do not yield to pedestrians and pedestrians have to choose safe gap on their own. This results in complex interactions between pedestrians and vehicles. Analyzing the interactions is important from both safety and performance evaluation of pedestrian crossings. The paper reports the applicability and generality of driver's gap acceptance models to the pedestrian's gap acceptance. Temporal and spatial critical gaps are estimated using both deterministic (Raff's and Ashworth's method) as well as probabilistic approaches (Maximum Likelihood method and Logit method). The data collected using video camera at two uncontrolled marked mid-block crossings resulted in 1107 lag/gap observations. The analysis found that (a) temporal and spatial gaps follow lognormal distribution, (b) speed of the conflicting vehicle has significant effect on the spatial gap acceptance, (c) critical gap values by deterministic methods are smaller than those by probabilistic methods, (d) temporal and spatial critical gaps by different methods vary between 3.6-4.3 s and 60-73 m respectively. These values are much lower than the widely accepted values calculated using HCM 2010. The analysis can be used in developing methodologies for safety and level of service evaluation of uncontrolled pedestrian crossings. © 2016 Elsevier Ltd.