Using policy‐capturing and direct‐ranking methodologies, managers made salary‐raise decisions for 104 hypothetical employees. These employees were experimentally varied on five manipulated characteristics: (1) performance level, (2) performance consistency, (3) tenure, (4) current salary, and (5) external job offer. The main effects of these five characteristics accounted for an average of 77% of the variance in the managers' raise decisions. On the other hand, the interactions accounted for virtually no variance. Substantial individual differences were found among managers in the weights they attached to the five characteristics, even though all of the managers were from the same organization. Substantial differences were also observed in the relative importance of the five characteristics as determined by policy capturing and direct ranking. Future research and administrative practice implications are suggested, especially those regarding individual differences in attribute weights among man