OBJECTIVES: Patient controlled analgesia (PCA) with pump delivery is the mainstay of modern postoperative pain management, with often better pain control compared to competing methods. We performed a data mining study to identify the clinical and hospital infrastructure correlates of PCA use. METHODS: Patients older than 18 years having major operative procedures expected to require strong opioid based post-operative pain control were selected from the Premier Perspective database. Obstetric patients were excluded. Two random samples were selected: a training sample of N=21,782 and a validation sample of N=21,538. Factor analysis mapped the 75 observed explanatory variables, not related to post-operative pain method onto 17 independent factors. Patient segmentation was performed based on cluster analysis. RESULTS: Thirteen distinct clusters were identified each with distinguishing demographic, clinical, payor and hospital setting features. Percent of PCA use in each of the segments ranged from 3% to 38%. Six segments, accounting for 3.5%, 1.8%, 8.4%, 4.1%, 6.1% and 9.2% of the total population had PCA usage below 10%. Only three segments had PCA usage more than 20% with 4.7%, 14.5% and 7.8% of the population. Qualitative evaluation indicated that the primary factor determining whether a segment had low or high PCA use was the circumstance of admission: for urgent admissions PCA was lowest, while the highest PCA use was observed in those segments characterized by elective admissions. CONCLUSION: Use of PCA differs between segments in a heterogeneous cohort of post-operative pain patients. That the driver of this difference was admission source may suggest that time required to prepare PCA is an important factor in its actual use and that thus easier PCA methods may increase patients use of PCA.