Four hundred and nine patients age 30 and over with leukemia or lymphoma and a subset of 127 patients with myelocytic leukemia were compared with equal numbers of matched hospitalized control subjects and matched subscribers to the Kaiser-Permanente Medical Care Program (member controls). Prior use of phenylbutazone was somewhat more frequent in cases than in member controls only, but the difference was not statistically significant and the association appeared to be explained by an underlying relation of musculoskeletal disease to these lymphohematopoietic cancers. The association of musculoskeletal disease with myelocytic leukemia was present even when attention was restricted to the period at least two or five years before leukemia diagnosis. There was no clear association of amount and duration of phenylbutazone therapy with risk of leukemia. Several other drugs used for musculoskeletal conditions showed positive or negative relations to subsequent leukemia.