Based on a modification of Berglas and Jones' (1978) design, conditions of contingent and noncontingent success and failure were manipulated to determine when and why individuals choose to adopt self-handicapping strategies. Male undergraduates (N=76) were informed that they were participating in a study investigating the effects of music on intellectual and creative tests. Contingency was determined by informing subjects accurately about their performance on 32 moderately difficult analogies while noncontingency was determined by informing subjects randomly about their performance on 24 of the 32 analogies. Feedback was manipulated by informing subjects they scored in the 95th percentile (success) or in the 25th percentile (failure). The major dependent variable was the selection of either performance-facilitating or inhibiting music to be listened to prior to an alleged retest. The selection of the inhibiting music was considered to be the manifestation of self-handicapping. Measures of control, attributions and expectations were used as ancillary dependent measures while the Multiple Affect Adjective Check List and the Multidimensional Internal-External Scale were used as premeasures. It was found that the best predictor of self-handicapping was reported feelings of depression, regardless of the experimental manipulations. Self-handicappers appeared to maintain an illusory sense of control on the experimental task, while paradoxically downplaying the overall importance of control in their lives. (Author)