This paper presents an analytic framework to measure the spatial segregation caused by reducing or forbidding the free movement of pedestrians, due to the existence of a highway or other type of transport facility with barriers that prevent pedestrians from crossing it. First, using empirical data from Berlin, London, Sydney and Santiago, it is shown that the proportion of walking as a function of travel distance approximately follows an exponential distribution. Then, probabilities of walking and expected walking distances are calculated under two alternative configurations -free vs constrained pedestrian crossing. Assuming an exponential distribution, we find that average walking distance increases by L/2 plus any extra walking distance due to the crossing itself (e.g., stairs, accessways to pedestrian overpasses), when pedestrian crossing is forced to be made every L metres. The model is applied in Santiago, on a road where a normal avenue was replaced by a segregated highway with pedestrian overpasses in specific locations to allow crossing. We show that the segregated facility decreases the probability of walking to places where walking distance has increased, worsening car dependency even for short trips. The greatest inconvenience is for people living directly adjacent to the highway, whose walking distance to cross the road is tripled on average. This is an estimation of the barrier effect produced by this type of segregated transport infrastructure. (C) 2014 Elsevier Ltd. All rights reserved.