Fifty-two adults who had previously received 4 x 20 micrograms doses of hepatitis B [Engerix-B] vaccine (appropriately administered into the deltoid muscle) and who had failed to develop detectable anti-HBs were randomized to receive a fifth dose of Engerix-B (either 20 micrograms or 40 micrograms) intramuscularly (deltoid). The participants were blinded as to the contents of the syringe. Anti-HBs was tested (by EIA) 3 months after the injection. Anti-HBs results from 45 non-responders were evaluable. Seven vaccinees were excluded; four of these on the basis of failure to have follow-up blood collected and three who were found to be anti-HBc positive (one HBsAg positive). Twelve of 22 (54.5%) of those receiving 20 micrograms of HB vaccine developed anti-HBs, whereas 10 of 23 (43.5%) who received 40 micrograms developed anti-HBs, showing no significant difference between the regimens. The mean geometric titres were 93 +/- 50 and 86 +/- 51 IU l-1, respectively. Vaccinee groups were well matched for age, sex and body mass index and the interval between injection and bleeding. Side-effects in those receiving the double (40 micrograms) dose were no different from those receiving the normal (20 micrograms) adult dose. On the basis of this study, a fifth dose of vaccine in non-responsive vaccinees is recommended. No significant advantage of 40 micrograms over 20 micrograms of vaccine was observed. Whilst smoking and obesity were common in this cohort of non-responders and probably contributed to the individuals primary non-responsive state, these factors had no unfavourable effect on response to a fifth dose of vaccine.