The avoidance of cytotoxic effects associated with protein misfolding has been proposed as a dominant constraint on the sequence evolution and molecular clock of highly expressed proteins. Recently, Leuenberger et al. developed an elegant experimental approach to measure protein thermal stability at the proteome scale. The collected data allow us to rigorously test the predictions of the misfolding avoidance hypothesis that highly expressed proteins have evolved to be more stable, and that maintaining thermodynamic stability significantly constrains their evolution. Notably, reanalysis of the Leuenberger et al. data across four different organisms reveals no substantial correlation between protein stability and protein abundance. Therefore, the key predictions of the misfolding toxicity and related hypotheses are not supported by available empirical data. The data also suggest that, regardless of protein expression, protein stability does not substantially affect the protein molecular clock across organisms.