One hundred and one patients on haemodialysis, 21 patients on peritoneal dialysis and 30 healthy controls received a trivalent split vaccine containing 15 micrograms haemagglutinin of a recent influenza A-H3N2, influenza A-H1N1 and influenza B strain, respectively. Antibody production after four weeks was determined by the haemagglutination-inhibition test and expressed as response rate, protection rate and overall mean fold increase. The patients on haemodialysis revealed a diminished seroresponse, as compared to patients on peritoneal dialysis and controls. For influenza A-H3N2, this was less distinct than for the other two antigens. In patients on haemodialysis the protection rate was 66% against the A-H3N2 vaccine component (versus 85% in controls, not significant), but only 25% against A-H1N1 and 27% against B (versus 84 and 77% in controls, p less than 0.001). Duration of haemodialysis up to eight years did not affect seroresponse. Patients on haemodialysis who were primed for influenza A-H1N1 in the period 1947-1957, reacted markedly better to the A-H1N1 vaccine component than subjects of other priming periods. A booster injection of the same vaccine dosage four weeks after the first immunization, performed in 98 patients on haemodialysis, was of little value: it had virtually no effect with regard to influenza A-H1N1 and influenza B, and showed, though significantly better, still poor results for A-H3N2. The differences in seroresponse between the A-H3N2 and A-H1N1 vaccine component suggest a major defect of primary, and a minor defect of secondary humoral response in patients on haemodialysis. The consequences for vaccine policy in these patients are discussed.