Medical diagnoses are often made on the basis of the presence of multiple symptoms. However, little is known about how the presence of multiple simultaneous symptoms may influence a bias in determining which symptoms are identified, in part due to a lack of an experimental analogue of this process. The current article presents a laboratory analogue of this process and explores whether over‐selectivity influences the ability to identify symptoms indicative of particular illnesses. In two experiments, participants completed a diagnosis task that required them to rate the degree to which symptoms predicted illnesses, with predictor symptoms being presented either singly or in compound. In both experiments, over‐selectivity was observed; one symptom of the compound received lower ratings, compared to the other element of the compound and the single predictor, while the other component received comparable ratings with the element. These findings are discussed in relation to associative accounts of over‐selectivity and as a procedure to study biases in medical decision making.Copyright © 2017 John Wiley & Sons, Ltd.