Misspecification tests of individual assumptions underlying regression models often lead to erroneous conclusions regarding source of misspecification. Monte Carlo experiments demonstrate that a comprehensive set of individual and joint tests reduces the likelihood of such conclusions. A practical testing strategy is proposed and suggestions made regarding its implementation.