The interaction between pedestrians and left-turning vehicles at signal- ized intersections are examined using the traffic conflict technique. Paramount was a comparison of the safety of left turns at two types intersections: T-intersections and X-intersections (cross-intersections). Previous research has indicated that T-intersections are more dangerous to pedestrians. In preparation for the comparison several traffic conflict definitions and their applications to pedestrians were evaluated. Use of a laptop computer for data collection was tested. Eight sites taken from intersections in Hamilton, Ontario, Canada, were selected. A conflict recording methodology was developed for T-intersections and X- intersections that consisted of recording data at various times along the paths of pedestrians and left-turning vehicles, and recording traffic con- flicts. Two computer programs were written for the data collection process: one for vehicles and one for pedestrians. Several statistical tests to relate traffic conflicts and the expected number of accidents were per- formed. These tests indicate that a positive correlation between traffic conflicts and expected number of accidents exists; they also suggest that T-intersections have a higher traffic conflict rate than X-intersections. The primary objectives of the described study are (a) to examine the interaction of left-turning vehicles and pedestrians at two types of signalized intersections using traffic conflicts; (b) to compare sev- eral traffic conflict techniques; and (c) to test the use of a laptop computer to record the traffic conflicts. Four sections describe the study. The first section reviews the lit- erature on pedestrian accidents and traffic conflicts, concentrating on accidents between left-turning vehicles and pedestrians, various traffic conflict definitions, and the validity of the traffic conflict technique. The second section describes the methodology developed to record traffic conflicts according to the different definitions and the data collection process. Results of the study are presented in the third section, which also examines the conclusions of Quaye et al. (1) and the relationship between conflicts and the expected number of accidents. The fourth section concludes with a discussion of the questions the study sought to answer. LITERATURE REVIEW A driver making a left turn at an intersection often has to keep track of several traffic elements simultaneously, including opposing traf- fic flow, traffic lights, and pedestrians crossing. According to vari- ous studies left-turning traffic generally constitutes about 20 percent of the approach traffic in urban areas; however, the proportion of accidents involving pedestrians and left-turning vehicles at inter- sections is slightly higher (20 to 30 percent of all pedestrian acci- dents in intersections), as shown in Table 1. TRANSPORTATION RESEARCH RECORD 1538 61 One-way intersections reduce the complexity of turning left by removing the opposing traffic flow, thus allowing drivers to con- centrate on the pedestrians in the crosswalk and the traffic lights. Two studies, one each conducted by Habib (2) and Fruin (3), have examined pedestrian accidents at signalized intersections on a one- way grid system (Manhattan, New York). They discovered that vehicle left turns were approximately four times more dangerous to pedestrians than through movements. In a more recent study, Almuina (4) examined accidents involv- ing left-turning vehicles and pedestrians at signalized intersections in Hamilton, Ontario, Canada, between 1983 and 1986. He found that about 32 percent of the pedestrian accidents involved left- turning vehicles. Almuina further analyzed pedestrian accidents by dividing intersections into three types: one-way–one-way, one- way–two-way, and two-way–two-way. He demonstrated that, with the exception of pedestrian accidents with straight-through vehicles, accidents involving left-turning vehicles had the highest proportion of accidents for all intersection types. Previous research has indicated that the most important factors in increasing the likelihood of pedestrian accidents are pedestrian and vehicle flows (5,6). Accordingly, numerous relationships between accident frequencies and traffic flows have been examined over the years. The most recent studies, such as Hauer et al. (7), assess the safety of intersections by using the product of vehicular traffic flows raised to a power. They suggest that other circumstances such as highway-rail grade crossings and accidents on two-lane highways support the “product-of-flows-to-power” relationship. Studies by Zegeer et al. (8), Hauer and Persaud (9), and Mengert (10) have reached similar conclusions. Quaye et al. (1) specifically examined the safety of pedestrians by using the product of pedestrian and vehicular flows raised to a power. Their research relied on accident and flow counts during 15-min periods from Monday to Friday for the years 1983 to 1986 in Hamilton. The multivariate accident prediction model was given as follows: Ê{m} = b 0  F 1 b 1  F 2 b 2 (1) where F 1 , F 2 = vehicle and pedestrian flows, respectively; b 0 , b 1 , b 2 = parameters to be estimated; m = entity (signalized intersection); E{m} = mean of such m’s for different intersections with flows F 1 and F 2 ; and ^ = estimate of E{m}. Because the safety of pedestrians at an intersection is influenced by its geometry, Quaye et al. separated fixed-cycle signalized inter- sections as well as the models into two categories. Analysis of Pedestrian Conflicts with Left-Turning Traffic DOMINIQUE LORD Transport Safety Group, Department of Civil Engineering, 35 St. George Street, University of Toronto, Toronto, Ontario, Canada M5S 1A4.