United States Department of Agriculture defines food insecure as answering affirmatively to three or more food insecurity questions describing a household's ability to acquire enough food. Households indicating low levels of food insecurity (one or two affirmative responses) are considered food secure. This paper compares the characteristics of households with one or two positive survey responses (termed marginally secure in this paper) to those with zero positive responses (food secure) and those with three or more positive responses (food insecure). The analysis utilizes Current Population Survey Food Security Supplement data to compare the characteristics and food purchasing of food secure, marginally secure and food insecure households using binomial and multinomial logistic regression and t-tests. Binomial logistic regression models indicate that grouping insecure and marginally secure households together does not change predictors of food insecurity. Multinomial logistic regression models suggest a three category definition of food insecurity is appropriate because there are distinctions among the three categories. There are significant differences in food spending across the groups. Prevalence of U.S. food insecurity and need for food assistance may be underestimated because marginally food secure households are considered food secure. The current measure fails to recognize that marginally secure households may experience poorer quality of life as do food insecure households.