To develop and validate a food frequency questionnaire (FFQ) for low-income, 1- to 3-year-old Hispanic, African-American, and white children. A convenience sample of low-income mothers provided dietary data for their child via FFQs and 3-day diet records. Participants (N=77) were recruited from Special Supplemental Program for Women, Infants, and Children clinics and public health clinics. All subjects were Hispanic, African-American, or white mothers of children 1 to 3 years old, 18 years of age or older, income less than 200% of the federal poverty level, and were literate in English. Wilcoxon signed-ranks tests were used for comparisons of food group servings on time 1 and time 2 FFQs and to examine differences in food group servings between the FFQ and diet records. Spearman correlations were calculated to assess test-retest reliability and the relative validity. Contingency tables were used to determine the degree of association between the FFQ and diet records. Reliability correlations were significant for all nine food categories (¯x=0.69), ranging from 0.53 (soups) to 0.84 (nonstarchy vegetables). Validity correlations were significant for all food groups (¯x=0.41) except starchy vegetables. Thirty-six percent of children were classified into the same quartile of food group intake and 78% into the same or within one quartile. The FFQ yielded excellent reliability and acceptable validity and can be used to assess food choices in a triethnic sample of low-income 1- to 3-year-old children.