This paper analyzes the relationship between output, employment, and physical and R&D capital, for a sample of 133 large U.S. firms covering the years 1966 through 1977. In the cross sectional dimension, there is a strong relationship between firm productivity and the level of its R&D investments. In the time dimension, using deviations from firm means as observations and unconstrained estimation, this relationship comes close to vanishing. This may be due, in part, to the increase in collinearity between trend, physical capital, and R&D capital in the within dimension, leaving little independent variability there. When the coefficients of the first two variables are constrained to reasonable values, the R&D coefficient is both sizeable and significant. The possibility of simultaneity between output and employment decisions in the short run is also investigated. Allowing for this via the use of a semi-reduced form equations system yields rather high estimates of the importance of R&D capital relative to physical capital. Our data do not allow us, however, to answer any detailed questions about the lag structure of the effects of R&D on productivity. These effects are apparently highly variable, both in timing and magnitude.