Criteria for assessing adequacy of cervicovaginal smears according to the Bethesda System include cellular composition and quantity, specimen preservation, fixation, and absence of obscuring elements. The reproducibility of adequacy assessment using these parameters was the focus of this study. Specimens were chosen to include a wide spectrum of clinical history and adequacy interpretations, but excluded cases with epithelial abnormalities and those judged "limited" because of insufficient transformation zone component. The 114 specimens were independently evaluated twice by five reviewers, with interceding randomization. Full interobserver concordance was obtained in 50.4% of cases, with consensus reached by three or more observers in 97.4% of cases. Of 2280 paired interpretations, 74% (1692) concurred. Pairwise kappa values for interobserver agreement ranged from .71 to .54 (very good to fair), with overall kappa = .61. Interobserver agreement for each category was near excellent for "satisfactory" (kappa = .73), good for "unsatisfactory" (kappa = .63), and only fair for "satisfactory but limited" (kappa = .48). Paired intraobserver reproducibility (agreement for two separate readings by the same observer) ranged from 69% to 77% (average, 74%), with good to fair kappa values (.66-.51). These findings indicate that the specimen adequacy guidelines of the Bethesda System provide good inter- and intraobserver reproducibility. The "satisfactory but limited" category appears to show the lowest concordance. Problem areas include evaluation of adequate minimum cellularity and estimation of number of cells obscured.