This study uses a natural experiment approach to evaluate the causal effect of three regular road pedestrianization schemes in Central Tokyo on traffic conditions in surrounding areas. Cross-sectional traffic flow data collected by the Metropolitan Police Department via traffic counters was used as the dependent variable and causal effect evaluation was done via a difference-in-differences (DID) approach, where the temporal pedestrianization of a street was used as the treatment of interest. We take advantage of (1) the variability in pedestrianization execution days and (2) the seasonal variability in pedestrianization time periods, to define the natural experiments. Furthermore, three different buffer areas of 500m, 750m and 1000m were used to evaluate the sensitivity of effect estimates to the definition of the impact area. Results suggest that irrespective of buffer size, the impact of temporal road pedestrianization on cross-sectional traffic flow of surrounding roads is negligible. Point estimates of treatment effect magnitudes across all models are small (ranging from − 4.8–6.9%) and estimated with high degrees of uncertainty, with most confidence intervals including zero. As such, results do not show evidence of practically important increases in traffic flow in the surrounding areas of pedestrianized roads and point to some degree of “traffic evaporation” likely resulting from drivers’ behavioral adaptation.