OBJECTIVE The aim of this study was to perform an external validation of 2 institutionally derived predictive models of laparoscopic conversion in colorectal surgery using the Mayo Clinic, Rochester (MCR) laparoscopic colon and rectal surgery experience. SUMMARY OF BACKGROUND DATA Two different predictive scoring systems of conversion in laparoscopic colorectal surgery were developed and published based upon single institution experiences. Neither model was validated on an independent data set. Thus, the utility of these models outside of their respective institutions is unknown. METHODS A prospectively collected data set of 998 laparoscopic colorectal procedures from MCR was analyzed. All patient-, procedure-, and surgeon-related factors used in both models were present in our data set. Logistic regression was used to evaluate their ability to predict conversion in our cohort. Model effectiveness was assessed by area under the curve from the logistic regression model, 95% confidence intervals for the observed number of conversions, and a goodness-of-fit test to compare the observed number of conversions with the predicted conversion rates for each score. RESULTS The cohort mean age of 552 women was 53, with a median body mass index of 25.2 kg/m. There were 382 right-sided, 251 left-sided, 46 rectal resections, and 151 proctocolectomies. Major diagnoses were inflammatory bowel disease 34%, cancer 18%, polyps 17%, and diverticular disease 13%. The overall MCR conversion rate was 15%. Several variables from the models were statistically significant predictors of conversion in our data set. However, both models performed similarly with an area under the curve of 0.62, suggesting that these models are of limited predictive value in our independent cohort with a performance closer to chance. The numbers of actual conversions were significantly different from the predicted number for both scoring systems. CONCLUSION Patient and clinical factors associated with laparoscopic conversion in colorectal surgery may be institution dependent. This finding cautions surgeons on the applicability of institution-based surgical predictive models. Independent data set validation is recommended before surgical predictive models are applied to general clinical practice.