This bulletin reports empirical findings on the determinants of food insecurity in the United States, using data from the 1989-91 Continuing Survey of Food Intake by Individuals and the 1992 Survey of Income and Program Participation. Descriptive statistics on food insufficiency status (a proxy measure for the most food-insecure households) are presented from both surveys. Multivariate logit models are used to study the effects of socio-economic characteristics on food insufficiency. Households with higher incomes, homeowners, households headed by a high school graduate, and elderly households were less likely to be food insufficient. Holding other factors constant, those in poverty were over 3.5 times more likely to be food insufficient. However, there was not a one-to-one correspondence between poverty and food insufficiency, since over 40 percent of food-insufficient households were not poor and about 10 percent of poor households were food insufficient. Food stamp benefit levels were inversely associated with food insufficiency.