OBJECTIVES: Bariatric surgery for the treatment of obesity and its underlying comorbid conditions is a well-tolerated and safe procedure. While a surgical approach is becoming the recommended method of treatment for morbid obesity, not all patients experience significant improvement. The objectives of this study were to utilize the GE Centricity electronic medical records (EMR) database to perform a retrospective cohort analysis on weight loss over time in patients with a record of having laparoscopic gastric banding or bypass surgery to identify predictors of weight loss following bariatric surgery. METHODS: Analyzable patient records were drawn from the GE Centricity database from January 1990 through March 2010. To be eligible for inclusion, patients had a recorded CPT Procedure Code of 43644/43645 or 43770; ≥18 years of age; and a valid recorded BMI within 6 months prior to procedure and at least one time point post surgery. Univariate and multivariable analyses were performed using SAS version 9.2. RESULTS: There were some notable pre-surgery differences between bypass and band cohorts, including age, weight, BMI, selected comorbid conditions and use of antidepressants.  Patients with pre-op BMI values of less than 30 did not experience sustained weight loss, on average, regardless of surgery type. Patients with pre-surgery BMI values of 30 to less than 35 fared better with banding (20+% BMI loss) than with bypass (5% BMI gain) at 2+ years post surgery. Finally, patients with pre-surgery BMI values of 35 and above had greater success, on the whole, with bypass surgery. CONCLUSIONS: The percent BMI loss over time by four cohorts of pre-surgery BMI show a trend that may be useful in predicting treatment success as defined by weight loss. This analysis of EMRs demonstrates potential clinical benefits when evaluating laparoscopic gastric bypass and banding in a real world setting.