We analyze the technology shocks identified by two different structural VAR models and compare them with classical and refined Solow residuals. The measures of technology are reasonably highly correlated. Alternative identifying restrictions in the VARs, however, have different theoretical as well as empirical consequences for the technology shocks. King et al.'s (1991) model and the classical Solow residual capture a mixture of technology and labor supply shocks, whereas the technology shocks from Galí's model and the refined Solow residuals are robust to the latter phenomenon. Moreover, we find that the two robust measures of technology are negatively correlated with hours worked.