In a recent study by Gelman&Markman (1986), boys and girls were found to show different patterns of inductive inference. Children were asked to infer whether an object (e.g. a shark) would be more likely to share a property with a dissimilar‐looking object from the same category (e.g. a tropical fish), or with a similar‐looking object from a different category (e.g. a dolphin). Boys made more category‐based inferences than predicted by chance for objects from biological categories like fish, but not for non‐biological objects such as gold. Girls showed the reverse pattern, making more category‐based inferences than predicted by chance for non‐biological objects, but not for biological items. The present study tested the possibility that the sex difference was due to the way the properties were described rather than the biological‐non‐biological difference in the items. In the Gelman&Markman study, the properties of biological items were presented as actions or states, while the non‐biological properties were presented as the consequences to the object due to an outside agent. The results of the present research indicated that the content of the items (biological versus non‐biological) is irrelevant; sex differences in patterns of correct inference are a function of the way items are described (actions/states versus consequences). Boys made significantly more correct inferences than chance for items whose properties were described as actions or states, while girls made significantly more correct inferences than chance for items whose properties were described as consequences. Thus, it appears that boys tend to use an object's actions as a basis for inference, while girls tend to use the consequences for an object when categorizing the object and making