Growing popular interest in positive psychology may have important implications for the measurement of well-being. Five studies tested the prediction that well-being ratings are influenced by desirability bias. In Study 1, participants (N=176) instructed to fake good endorsed higher well-being; those instructed to fake bad endorsed lower well-being, compared to controls. In Studies 2 and 3 (N’s=111, 121), control participants endorsed higher levels of well-being compared to those attached to a bogus pipeline. These differences were mediated by desirability bias. In Study 4 (N=417), instruction manipulations did not affect well-being levels, but presenting a desirability measure prior to well-being measures attenuated the correlations between them. In Study 5 (N=391), however, this order effect did not replicate. We discuss the importance of continued vigilance for desirability bias in well-being research as a ready solution to this clear problem remains elusive.