A randomized double-blind comparison of five killed hepatitis A vaccine preparations was carried out with eligible medical student and staff volunteers. Vaccines were prepared in M RC-5 cells and formalin-inactivated. Three monthly injections of 1 ml in the deltoid muscle were given. Group A received the CLF strain at a dose of 360 ELISA units (El.U) in 0.5 mg aluminium hydroxide (n = 35). The other groups received the HM175 strain as follows: 180 El.U in 1 mg aluminium hydroxide (to group B, n = 42), 360 El.U in 0.5 mg aluminium hydroxide (to group C, n = 40), 360 El.U in 1 mg aluminium hydroxide (to group D, n = 39) and 720 El.U in 1 mg aluminium hydroxide (to group E, n = 43). The geometric mean anti-HAV concentration (GMC) measured in mIV/ml by an ELISA method one month after each injection were: group A, 223, 480, 1635; group B, 123, 221, 649; group C, 185, 365, 1085; group D, 144, 323, 1076; group E, 229, 646, 2521. At month 6, the GMC had fallen by approximately 20%. Seroconversion as measured by ELISA was 100% in groups A and E after one injection, and 100% in all groups after three injections; after two injections, only one subject in group C was still negative. The dose effect with HM175 vaccine was significant. There was a good correlation between ELISA and neutralization (radioimmunofocus inhibition test) titres. One month after the second dose, all subjects in groups A and E had both hepatitis A virus immunoglobulin M (HAV IgM) geometric mean titre, (GMT > 5000) and IgG (GMT > 25,000) as measured by a sensitive terminal dilution ELISA.(ABSTRACT TRUNCATED AT 250 WORDS)