This study contributes to existing literature on firms’ innovative activity examining the relationship between the R&D rivalry and spillovers at the firm level. In particular, we present an empirical analysis in United States, Japan and Europe based upon a new dataset composed of 879 worldwide R&Dintensive firms. In order to identify the technological proximity, we use the Jaffe industry weight matrix, based on the construction of technological vectors for each firm, where its patents are distributed across technology classes, in such a way that we compute knowledge spillovers. Opportune econometric techniques, which deal with both firm’s unobserved heterogeneity and weak exogeneity of the explanatory variables, are implemented. In order to test the robustness of our results, we introduce also the combined spatial-autoregressive model with autoregressive disturbances and additional endogenous variables. The empirical results are differentiated across countries, and suggest that the spatial effects are statistically significant.