We partially test and extend Stone & Colella's (1996) model of factors affecting the treatment of persons with disabilities. We conducted a laboratory experiment designed to predict helping behavior toward a coworker with a disability on a reward-independent task. Data were collected in a survey designed to measure some of our predictors. The experiment included a confederate, in whom we manipulated the presence and type of a disability (no disability, mental disability, and physical disability) and 133 participants. We ran a manipulation check on 84 participants in a holdout sample. Results indicate that impression management, equity preference, and type of disability were significantly related to helping behavior, but presence of disability and feminine gender-role identity were not.