OBJECTIVES: Per member per month (PMPM) utilization rates are commonly reported in the medical literature to compare differences in costs and other outcomes across various health care technologies and interventions. A limitation of PMPM estimates is that a confidence limit around the point estimate is not obvious or available from standard statistical software. Our objective is to demonstrate various methods of calculating confidence intervals for PMPM utilization rates. METHODS: Several methods were used to estimate confidence intervals surrounding PMPM estimates including Fieller's method and Monte-Carlo (MC) simulation. Women with at least one prescription fill for alendronate, risedronate, or ibandronate during 2006 in a large managed care data set were used as a sample to generate PMPM estimates and 95% confidence intervals for bisphosphonate drug cost, all hospitalization cost, hospital days, and number of hospital admissions during the calendar year of 2006. RESULTS: There were 34,675 women in our sample. The PMPM estimate of bisphosphonate drug cost was $23.48. The 95% confidence intervals generated by the Fieller and MC methods were ($23.21, $23.75) and ($23.45, $23.87), respectively. The PMPM hospitalization cost was $242.28: Fieller and MC 95% confidence intervals were ($221.53, $263.03) and ($227.74, $259.99), respectively. The PMPM estimate of hospital days was 0.108 days: Fieller and MC 95% confidence intervals were (0.098, 0.118) and (0.100, 0.116), respectively. The PMPM point estimate for number of hospital admissions was 0.0137: Fieller and MC 95% confidence intervals were (0.0131, 0.0142) and (0.0133, 0.0142), respectively. CONCLUSION: The Fieller and MC simulation methods produced similar confidence intervals for PMPM estimates for each of the outcomes of interest. Use of these methods would improve the utility of PMPM point estimates in comparing health care technologies.