During the National Influenza Immunization Program in 1976, 147 children with neoplastic diseases received Wyeth split-product bivalent influenza vaccine: A/New Jersey/8/76 (HSW1N1), A/Victoria/3/75 (H3N2). Thirteen normal siblings served as controls. Seventy-one patients received two doses of the vaccine four weeks apart. After the second injection of A/NJ/8/76, there was a difference between the response of the patients on chemotherapy and those off therapy greater than or equal to 30 days--38% vs. 76%, P less than 0.01 for four-fold rise and 26% vs. 57%, P less than 0.05 for the attainment of protective (greater than or equal to 32) hemagglutination inhibition (HI) titers. These differences were observed in both leukemia-lymphoma and solid tumor patients. There was a difference in HI titers to A/Vic/75 between patients on and off chemotherapy after a single injection, 34% vs. 71%, P less than 0.001 for a four-fold rise. After the second immunization, only 52% on, and 86% off therapy (P less than 0.05) had a four-fold rise in titers. Thirty-two percent of the patients on treatment who achieved "protective" titers did so only after the second immunization. Immunoglobulin levels and neutropenia did not correlate with the inability to obtain a four-fold rise in titers. Our findings suggest that patients on chemotherapy cannot be effectively vaccinated by a new antigen, and that single yearly boosters may be insufficient for recall of old antigens. Patients off chemotherapy greater than or equal to 30 days respond as normal controls.