The present experiments examined the relative effects of performance goals (Locke & Latham, 1990), directed at increasing measurable performance outcomes, and mastery goals (Dweck, 1986), directed at increasing ability, on simple and novel, complex tasks, and on complex tasks over time. The hypotheses that task complexity would moderate the relationships between goal direction and goal specificity on novel, complex task performance were not supported. Non-specific mastery goals did not increase performance over difficult, specific performance goals on novel complex tasks. The hypothesis that previous goals would interact with subsequent goals on performance on complex tasks was supported, however. Participants who had been given a mastery goal in the first session did better when given a performance goal in the second session, and participants who had been given a performance goal in the first session did better when given a mastery goal in the second session. Limitations of the studies and implications are discussed