Guided imagery, as other nonpharmacologic strategies, has been demonstrated to be useful for some patients. However, no tested method exists to identify which patients are likely to benefit from this pain management strategy. This pilot study tested a model to predict success with guided imagery. Major concepts tested included imaging ability, outcome expectancy, history of imagery use, match with preferred coping style, and perceived credibility of the imagery provider. A one-group pretest-posttest design was used. A sample of 62 hospitalized cancer patients currently experiencing pain rated >/=3 on a 0 to 10 scale completed questionnaires and used an audiotaped imagery intervention. Pain outcomes examined included mean pain intensity and distress, positive and negative affect, and perceived control over pain. A path analysis was conducted using multiple regression to evaluate relationships proposed in the model. Previous history with imagery predicted outcome expectancy. Imaging ability predicted mean pain intensity, positive affect, and perceived control over pain. Outcome expectancy was not a significant predictor of any pain outcomes. Baseline status and concurrent symptoms, measured as covariates, also played a significant role in predicting outcomes. Variance explained in pain outcomes ranged from 10% to 52% (adjusted R(2) = 3% to 48%). Further exploration of model variables is warranted. Findings suggest that after considering current symptom experience, imaging ability may be a useful variable to assess in order to determine whether guided imagery is an appropriate intervention for individual patients.