AbstractOver 180,000 black soldiers served in the Union Army in the American Civil War. They endured horrific deprivation and disease, and also substantial institutional discrimination in such areas as wage and promotion. Nevertheless, most never deserted their duties as soldiers. However, it is not clear what distinguished those who did desert from their counterparts. The present case‐control study fills this gap. It applies social bonds theory to explain desertion among blacks. The general hypothesis is the greater the bonds or stakes to the military unit, the lower the risk of desertion. Data refer to the First Michigan Colored Infantry (First Michigan) and are from the Bentley Historical Library, University of Michigan. They refer to the population of all 189 deserters in the First Michigan and a comparison group of non‐deserters. Available measures of social bonds include substitute status (substitutes received a bounty upon completion of service), receiving a wound (having “skin in the game”), and noncommissioned officer status. In addition, loyalty to family is framed as a risk factor. The dependent variable is desertion, a dichotomy (0,1). Results are adjusted for other measurable variables. A multivariate logistic regression analysis determined that measures of military social bonds were protective factors against desertion. Substitutes were 73% less apt, wounded soldiers 93% less apt, and noncommissioned officers were 79% less apt to desert than their counterparts. However, family bonds were a risk factor. The model correctly classified 77.95% of the cases. The results largely support a social bonds theory of deviance, but also may support views of the conflicting bonds between family and state as increasing desertion. The investigation is the first quantitative study of desertion among blacks in the Civil War.