In this paper, a numerical model was used based on the tool-piece geometric intersection, which enables the determination of a roughness profile as a function of feed, tool radii and tool eccentricity in side milling processes. Average roughness R-a and peak-to-valley roughness R-t were obtained for a group or family of tools, defined by an average radius value and a standard deviation value of all tool radii. The Monte Carlo method was used to generate N random combinations of radius values for each family, according to a normal distribution. The model was validated by means of experimental tests. For each family of tools, roughness distributions were obtained at different feed values and for different eccentricity values. It was found that the higher eccentricity, the more asymmetrical roughness distributions become.