The food stamp program (SNAP) is one of the most important elements of the social safety net and is the second largest anti-poverty program for children in the U.S. (only the EITC raises more children above poverty). The program varies little across states and over time, which creates challenges for quasi-experimental evaluation. Notably, SNAP benefit levels are fixed across 48 states; but local food prices vary widely, leading to substantial variation in the real value of SNAP benefits. In this study, we leverage time variation in the real value of the SNAP benefit across markets to examine the effects of SNAP on child health. We link panel data on regional food prices and the cost of the Thrifty Food Plan [TFP], as measured by the USDA's Quarterly Food at Home Price Database, to restricted-access geo-located National Health Interview Survey data on samples of SNAP-recipient and SNAP-eligible children. We estimate the relationship between the real value of SNAP benefits (i.e., the ratio of the SNAP maximum benefit to the TFP price faced by a household) and children's health and health care utilization, in a fixed effects framework that controls for a number of individual-level and region characteristics, including non-food prices. Our findings indicate that children in market regions with a lower real value of SNAP benefits utilize significantly less health care, and may utilize emergency room care at increased rates. Lower real SNAP benefits also lead to an increase in school absences but we find no effect on reported health status.