Many questions face researchers conducting personality disorder research. Which measure should be used? When should patients be evaluated--at initial presentation when they are still symptomatic, or after symptom resolution? Can a patient accurately describe his or her personality, or must an informant be contacted? What do you do when personality traits change during an individual's lifetime? I discuss these and other issues of personality disorder assessment in the DSM-III/DSM-III-R era, and examine the empiric literature bearing on these questions. First, I review studies of diagnostic reliability and stability, because studies of the above questions can only be interpreted in the context of the limitations imposed by modest reliability.