Background. Positive deviance is increasingly employed in international development activities to permit the utilization of proven local solutions. Including positive deviance methods in evaluation analysis, particularly in places like Bihar, India, where the rates of child underweight hover at 55%, can help identify project activities and household characteristics that affect key outcomes. These can, in turn, inform decision-making regarding the intensification of particularly promising activities. Objectives. To apply positive deviance analysis to the Dular program in Bihar, a community-based nutrition program that seeks to improve the impact of India’s Integrated Child Development Services on young children. Methods. In order to assure that desired program outcomes were not dependent on higher economic status, the analysis isolated a subset of program beneficiaries—the poorest children with the best nutritional outcomes—and examined the behavioral and project factors that may have brought about positive results in this subgroup. The data for this analysis were drawn from a 2005 program evaluation with a sample of 1,560 children. Results. The analysis found that positive deviant children with normal nutritional status in the poorest 50% of Dular households were introduced to complementary food almost 2 months earlier (7.18 vs. 9.02 months of age) than severely malnourished children, were more than twice as likely to use soap for handwashing after defecation (25.0% vs. 11.8%), and were more than seven times as likely to have literate mothers (25.0% vs. 3.5%). Conclusions. The analysis suggests that programmatic efforts relating to these activities have been particularly effective and may well deserve increased investment.