BACKGROUND: There are many reasons why an investigator may choose to share results with survey respondents; however, the practice is not universal. Moreover, while there is some evidence that the practice increases response rates, it is of limited generalizability. OBJECTIVE: The aim of this study was to determine both the impact on survey response rates of offering study results to participants, and the extent to which surveyed individuals with diabetes mellitus want results reported back from a postal questionnaire. METHODS: In November-December 2009, we surveyed 4796 randomly selected individuals from a US Midwestern health system diabetes registry. Half of the sample was randomized to receive an offer of survey results as an incentive. Response rates were compared across the two groups. Within the group offered results, we compared those that requested them with those that declined with respect to demographic characteristics, self-reported diabetes severity, and characterization of care received for diabetes. RESULTS: The overall response rate was 42.8% (n = 2055). Offering survey results did not impact response rates overall or within specific subpopulations. Of those randomized to the offer of results, 62.1% requested the results (n = 618). Requests for results differed by educational attainment and the rating of diabetes care as poor, although only education remained a significant predictor in multivariate analysis. There were no differences in rates of requests by disease severity or overall health status. CONCLUSION: Offering survey results to potential respondents did not increase the likelihood of response among a patient population with diabetes. Of those that were offered results and returned the survey, 62.1% requested that they receive study results, indicating a potentially important educational opportunity. However, as result requests were driven more by educational attainment than by health status, investigators may want to send results to everyone, not just those that request them.