To estimate the effect of the US Supplemental Nutrition Assistance Program (SNAP) on the food security (consistent access to adequate food) of recipients, net of the effect of the self-selection of more food-needy households into the programme.

The food security of current SNAP recipients and recent leavers is compared in cross-sectional survey data, adjusting for economic and demographic differences using multivariate logistic regression methods. A similar analysis in 2-year longitudinal panels provides additional control for selection on unobserved variables based on food security status in the previous year.

Household survey data collected for the US Department of Agriculture by the US Census Bureau.

Households interviewed in the Current Population Survey Food Security Supplements from 2001 to 2009.

The odds of very low food security among households that continued on SNAP through the end of a survey year were 28 % lower than among those that left SNAP prior to the 30-d period during which food security was assessed. In 2-year panels with controls for the severity of food insecurity in the previous year, the difference in odds was 45 %.

The results are consistent with, or somewhat higher than, the estimates from the strongest previous research designs and suggest that the ameliorative effect of SNAP on very low food security is in the range of 20–50 %.