This paper provides evidence of a problem with the influential testing and assessment of Solow’s (1956) growth model proposed by Mankiw et al. (1992). It is shown that when the assumption of a common rate of technical progress is relaxed in the neoclassical model, the goodness of fit of Mankiw et al.’s equation improves dramatically. However, and more importantly, it is shown that this result, as well as the magnitude of estimates obtained, merely reflects a statistical artifact. This has serious implications for the possibility of actually testing Solow’s growth model.