By early 2011, nearly one out of every eight individuals in the US received benefits through the Supplemental Nutrition Assistance Program (SNAP). By definition, SNAP recipients have low earnings during program participation, but less is known about the long-term earnings patterns of people who have received benefits from the program. In this paper I use Social Security Administration administrative earnings data linked to the Survey of Income and Program Participation (SIPP) to assess the effect of SNAP participation on long-term earnings. To do so, I observe complete spells of SNAP participation in the SIPP and then track the earnings of participants before and after those spells. The administrative data allows me to follow earnings back to 1951 and for about 10 years after each program spell. The estimates should provide insight as to whether the program acts as a welfare magnet or as a work support for people experiencing negative transitory shocks to their labor force participation and earnings receipt. If earnings after a SNAP spell are significantly lower than before participation, it would suggest the program acts as a welfare magnet. If earnings after SNAP participation return to their pre-SNAP spell levels, that would suggest the program acts as a work support.