We performed a double-blind, randomized trial to compare the immunogenicity and reactogenicity of four conjugate Haemophilus influenzae type b vaccines given to infants 2, 4, and 6 months of age. Adverse reactions attributable to the vaccines were few and minor. The rates of systemic reactions did not differ among the various vaccines and were similar to those seen among children receiving conventional diphtheria-tetanus-pertussis vaccine. However, the four conjugate H. influenzae type b vaccines differed markedly in ability to stimulate antibody production. Mean antibody levels after three injections of polyribosylribitol phosphate conjugated with mutant diphtheria protein (PRP-CRM) or polyribosylribitol phosphate conjugated with tetanus toxoid (PRP-T) were 3.08 micrograms/ml and 3.64 micrograms/ml, respectively, significantly higher than those after the use of polyribosylribitol phosphate conjugated with outer-membrane protein of Neisseria meningitidis (PRP-OMP) (1.14 micrograms/ml) or polyribosylribitol phosphate conjugated with diphtheria toxoid (PRP-D) (0.28 microgram/ml). Only PRP-OMP produced a clinically pertinent elevation in antibody level after two injections (0.84 microgram/ml); the third injection of PRP-OMP produced a modest but statistically significant further elevation in mean antibody level (1.14 micrograms/ml). Only 29% of infants receiving PRP-D had antibody levels of 1 micrograms/ml, compared with 55%, 75%, and 83% of those receiving PRP-OMP, PRP-CRM, and PRP-T, respectively. We conclude that all four vaccines are safe and that all but PRP-D appear appropriate for use in a primary immunization series during infancy. The unique serologic response to PRP-OMP offers both advantages and disadvantages in comparison with PRP-CRM and PRP-T.