At the Fifth International Workshop on Railway Noise in 1995, two papers reported apparently con#icting results relating to the e!ects of transverse pro"les on noise. Remington and Webb reported theoretical results showing that conforming pro"les could lead to a reduction in noise generation. Dings and Dittrich reported experimental results which showed apparently inconsistent roughness-to-noise behaviour between wheels with di!erent braking systems. They suggested that these results could be explained by an increase in noise for a given roughness level when the pro"les are conforming. This paper describes a new analytical investigation of the e!ects of transverse pro"les which aims to resolve this con#ict. It is based on new very detailed roughness measurements. The data were analyzed "rst using a discrete point reacting spring model (DPRS). The transverse pro"les of the wheel and rail were replaced with circular transverse pro"les of di!erent radii of curvature. It was found that the average roughness across the contact zone decreased as the width of this zone was increased, but that these reductions were only modest. The DPRS predictions showed little or no decrease in interaction force with increasingly conforming pro"les. The TWINS model has been used to perform calculations which con"rmed that, for the roughness data examined, little noise reduction bene"t could be expected from a conforming wheel, and in some circumstances it could lead to an increase in noise. TWINS was also used to study the e!ects of the actual transverse pro"les in an attempt to explain more fully the measured results. Some useful conclusions are drawn, but a full explanation of the measured results has not yet been obtained as all three types of braking system lead to some degree of conforming pro"les.