Pedestrian walking distances in a central business district are mostly dependent on the arrival mode in that CBD/central business district and the layout of the transportation network on which these modes operate. Changes to the existing pattern of operations, in terms of the location of new transportation facilities or relocation of existing facilities, can have a significant impact on walking distances. However, often the prime concern has been the impact of these changes on the flow of traffic, with little or no emphasis on the pedestrians. As opposed to arbitrarily derived acceptable walking distances suggested by some researchers, an approach based on findings from a series of surveys conducted in Calgary, Alberta, Canada, is proposed. A set of characteristics that influence the distribution of walking distances is identified. Depending on the form of the distribution, the "critical" distance that would be acceptable to a group of people can be derived using sinple calculus. The critical distance will be the distance at which the rate of change of the slope of the frequencey distribution of the walking distance for that group of people is greatest. Since the walking distance distributions indicate the propensity to walk, they indirectly represent the actual feelings of the people. This approach also facilitates decisions regarding locations, evern where information regarding walking habits is scarce or unavailable. The application of Bayesian statistical decision theory enables one to estimate the appropriate distribution and the critical distance associated with it.