People in fragile environments face various shocks that negatively affect their nutrition. Many governments put policy mechanisms in place to promote recovery of households after adverse shocks; however, resilience is difficult to measure because some apparent recovery could be the result of statistical randomness and reversion to trends. This paper demonstrates a new approach to measuring nutritional resilience in a population. As our starting point, we use the common definition of resilience as ‘recovery after decline’, but also require that the degree of recovery should exceed stochastic expectations. Using maternal and child nutrition data from Nepal, Bangladesh, and Uganda, we find that observed recovery is not always statistically significant and does not always satisfy this definition of resilience. We identify household and community factors that are correlated with measurable nutritional resilience, and recommend points of entry for policies designed to enhance resilience in resource-constrained settings.