To determine the sensitivity and specificity of the periodic acid-Schiff (PAS) stain in the diagnosis of acute leukemia in light of the finer characterization of this disorder now available through immunophenotyping, we examined the blasts from 51 patients with newly diagnosed acute leukemia by morphological, cytochemical, and immunophenotypic analyses. The 51 patients represented every new case of acute leukemia subjected to cytochemical stains and flow cytometry between July 1987 and February 1989. By cell-surface marker analysis, 29 exhibited lymphocytic lineage, while 21 were myelocytic. One was mixed lineage. The PAS positivity, defined by the presence of blocks or coarse granules in 5% or more of the blasts, was found in 15 of 29 lymphoblastic leukemias and in four of the myeloblastic leukemias. However, PAS-positive lymphoblastic leukemias were negative with the other cytochemical stains: myeloperoxidase, Sudan black B, and alpha-naphthyl butyrate esterase. The PAS-positive myeloblastic leukemias were positive with at least one other stain. Three cases of myeloblastic leukemia exhibited greater than 10% PAS-positive blasts, with all three being acute monoblastic leukemia. Thus, the sensitivity and specificity of the PAS stain alone for lymphoblastic leukemia was 52% (15 true positives of 29) and 81% (four false positives), respectively. The sensitivity of a cytochemical-staining combination of PAS positivity and myeloperoxidase, Sudan black B, and alpha-naphthyl butyrate esterase negativity in defining cases of lymphoblastic leukemia remained at 52%; however, the specificity of this combination for lymphoblastic leukemia was 100% (no false positives). Thus, a positive PAS stain, in combination with negative myeloperoxidase, Sudan black B, and alpha-naphthyl butyrate esterase stains, continues to have a diagnostic role in the distinction between lymphoblastic and myeloblastic leukemia, and greater immunologic sophistication serves to support this position.