Over the past two decades there have been several empirical studies of the diffusion of industrial innovations. The work of Carter and Williams (1957, 1958), Mansfield (1968), Mansfield, et al. (1971), Metcalfe (1970), Hsia (1973), Nabseth (1973), Ray (1969) and others has generated much evidence about the nature of the diffusion process and about the determinants of the rate at which it occurs. However, it seems that in many respects we still know very little. One area where this is perhaps most obvious is in understanding differences in the way different industries react to innovations. In a recent study the author (Romeo, 1975) attempted to deal explicitly with this problem by examining interindustry differences in the rate at which a capital-embodied process innovation displaced older methods. Specifically, he examined the rate at which numerically controlled machine tools (NC) replaced conventional machine tools in ten US industries' new machine tool purchases. Although such a displacement process had not been expressly considered before, it was found that Mansfield's (1968) widely used model of the imitative process could be usefully applied. Furthermore, the interindustry differences were pronounced and were in large part explainable by a number of characteristics of the innovation and the industries. In this paper, we again consider the diffusion of numerically controlled machine tools. However, the rate of diffusion studied is a more "traditional" one, the rate of imitation-the rate at which first use of an innovation spreads from firm to firm in an industry. In the case of NC this rate of imitation, which is measured as a rate of increase in the percentage of firms using NC (regardless of their extent of use) is not significantly correlated across industries (r = 0.193) with the aforementioned rate of displacement. The goal of the work here is threefold: first, to determine the existence of any interindustry differences in this rate; second, to,explain determinants of these differences and third, to check the dependency of these results on the particular measure of the rate of diffusion employed.