The Psychiatric Status Schedule (PSS) is a widely used interview that was designed to improve the research value of clinical judgments. Although it was developed with psychiatric patients, its authors hoped it could be used to evaluate nonpatients, a capability that would make it a much needed tool for epidemiologic research. The present study tests the internal consistency reliability of scales drawn from the PSS in both a general population sample (n. = 133) and a patient sample (n. = 100). Like the PSS's authors, we found a wide variety of clinically meaningful scales reliable for use with patients. In striking contrast, however, most of these proved unreliable in the general population sample. Speculative explanations are offered for the failure of most of the PSS scales in the general population sample and for the success of a few.