Most pedestrian accidents in built-up areas occur at intersections. Even after signalization the number of accidents involving pedestrians often remains high. After reviewing the published evidence, this paper describes how the Traffic Conflicts Techniques has been used to examine the risk to pedestrians at 120 intersections. The principal results indicate that signalization of a high-speed intersection (mean speed above 30 km/h in at least one arm) reduces pedestrian risk to approximately half. If mean speed in every arm is below 30 km/h signalization also reduces pedestrian risk, as long as most vehicles are not turning. Another finding from these studies was that a crosswalk should be located less than two meters from the intersection to optimize pedestrian safety. The conflict studies as well as analyses of accident data show that one should examine separately accidents between turning vehicles and "green-walking" pedestrians and accidents involving pedestrians walking against red light. The effect of an exclusive pedestrian signal phase (scramble) was tested at three sites and proved to be very safety-beneficial in a small town, while in Stockholm it did not prove effective because of a high percentage of red-walkers. Vehicle delay, as well as pedestrian delay, increased at all three sites. Data collected at 152 crosswalks has been used to estimate the parameters of a multivariate model of the frequency of "red-walking." The size of the town and traffic volumes appear to be the major factors influencing this frequency. Additional insight has been obtained from personal interviews of 450 persons. These indicate that shorter waiting times and police enforcement are considered the most efficient measures to reduce the frequency of red-walking.