To test the validity of the seven Diagnostic Statistical Manual-IV (DSM-IV) dependence (formerly termed 'addiction') criteria as applied to eating behavior, I administered a paper and pencil adaptation of the criteria to 48 members of a church attending a health fair and 17 students attending an adult education class on the effects of refined carbohydrates. Participants endorsed criteria meaning they indicated true or yes to the behavior described in a criterion without the evaluation of a trained interviewer. Two adaptations were developed to test for convergent validity. The first was a minimal rewording of the DSM-IV criteria. The second was more descriptive of symptoms and was adapted from a validated instrument, the Structured Clinical Interview for the DSM (SCID). Answers to the seven criteria were weighted for the number of foods indicated as used in the behavior. The test was given twice to evaluate for test-retest reliability (n=13). BMI and physical activity data were tested for predictive validity. Principal component analysis was performed as a pilot. Cronbach's alpha for participants taking the test for the first time (excluding the population who took the test a second time) was .67 for the DSM adaptation and .81 for the SCID adaptation. Pearson's Correlation for the test-retest was .60 (p=.015) for the DSM and .51 (p=.05) for the SCID. Convergent validity was .517 (p=.00). Discriminant validity using categorical analysis was demonstrated for BMI (p<0.0001). Discriminant validity was weakly demonstrated for physical activity but did not reach significance (p=0.20) Principal component analysis did not reveal redundancy. The DSM-IV dependence criteria performed well when applied to eating behavior. The moderate success of test-retest may be the result of test fatigue or of having a week to think about and observe eating behaviors described in the criteria. The moderate convergent validity seemed to be due to vagueness in DSM adaptation versus more detailed descriptions in the SCID. The low significance for discriminant validity may be due to outliers. These results demonstrate the possibility that overeating may be a substance use disorder ('addiction'). If further research should confirm this finding, it could represent an important breakthrough in the understanding, prevention, and treatment of obesity.