A 'double-blind', randomized, placebo controlled study of an oral inactivated whole cell plus recombinant B subunit (WC/rBS) cholera vaccine was conducted during February-March 1992 in Peru in 346 military recruits, 307 (89%) of whom received 2 oral doses of vaccine or Escherichia coli K12 placebo, 2 weeks apart. Paired serum samples were obtained from 155 (50%) of the recipients of 2 doses. An epidemic of cholera took place between doses. No difference in cholera attack rates was detected between vaccine and placebo recipients after one dose (8% versus 14%). Seroconversion (4-fold or higher increase in vibriocidal antibody titres) was detected in 90% and 80% of vaccine and placebo recipients, respectively, with low pre-existing vibriocidal titres (< 0.01). The anti-cholera toxin seroconversion rate among those with low pre-existing titres was higher in vaccinated subjects (97%) than in placebo recipients (68%) (P < 0.01). Administration of 2 doses of WC/rBS vaccine concomitantly with natural V. cholerae O1 infection enhanced the serum anti-cholera toxin response. The immune response to the whole cell component of the vaccine was reduced by high pre-existing vibriocidal antibody titres.