In prior attribution research on Kelley's covariance theory, researchers have generally accepted the person-stimulus-circumstance attribution taxonomy based upon covariation between contextual variables (e.g., consistency) and “global” measures of attributions. This article examines the person-stimulus-circumstance taxonomy using both specific-item attribution measures and the more commonly used general-item measures. Factor-analysis results support the person-stimulus-circumstance typology. However, convergent and discriminant analyses indicate that different concepts are being measured by the specific- and general-item measures when the research moves from one level of abstraction to another.