This paper evaluates the causal impact of the Jubilee Line Extension (JLE) in London in 1999 on air quality, considering the spatial heterogeneity of impacts. Controlled on seasonality, trend, autocorrelation, and meteorological factors, a segmented regression model is specified for each pollutant concentration time series in each buffering area. By specifying the segments with the start of interruption, it can be treated as a sharp discontinuity design in time. The total effect is the sum of the impact from the current period and the stacked impacts from the lag periods. The selection of seasonality and autocorrelation variables varies with the pattern of each time series. Results are robust in general. Heterogeneity of impacts is evident across spatial and monitoring locations. It shows that the JLE increased NO2 and NOx background concentration by 4-13% and CO outside 2.5 km by 8-9% within one year after JLE, but decreased the CO roadside concentration within 2.5 km by 8%. This implies that an increase in urban transit supply may not necessarily lead to an improvement in air quality, where the induced economic activity and land use redistribution can be potential reasons.