Though many studies have measured patient satisfaction with pain management using the American Pain Society (APS) Satisfaction Survey or its variants, little is known about the relationship among the survey items, or whether items relate to satisfaction at all. In an effort to refine the measurement of patient satisfaction, a modified version of the APS survey, which was given to 787 patients as part of a study of postoperative pain management in six community hospitals, was subjected to principal components analysis to determine the survey's empirical structure. Correlations among the five components found were low; a weak relationship (r = -0.24) was discovered between pain intensity and satisfaction. A heuristic model estimated by structural equations analysis yielded additional insights. Though many items thought to influence patient satisfaction were not closely related to patient-reported satisfaction, they indicate important clinical factors relevant to quality of care, and thus, to continuing quality improvement (CQI) efforts. Results suggest that satisfaction was influenced by effectiveness of medication, independent of pain intensity, and by communication. Pain severity ratings near the time satisfaction was measured were more influential than earlier ratings.