Do children expect an expert in one domain to also be an expert in an unrelated domain? In Study 1, 32 three- and four-year-olds learned that one informant was an expert about dogs relative to another informant. When presented with pictures of new dogs or of artifacts, children who could remember which informant was the dog expert preferred her over the novice as an informant about the names of dogs, but they had no preference when the informants presented artifact labels. In Study 2, 32 children learned that one informant was incompetent about dogs whereas another was neutral. In this case, children preferred the neutral speaker over the incompetent one about both dogs and artifacts. Taken together, these results suggest that for children, expertise is not subject to a "halo effect," but incompetence may be subject to a "pitchfork effect".