Children aged 7-14 years in Novgorod, Russia, were given Russian live cold-adapted or inactivated influenza vaccines or placebo over a 2-year period. Schools were randomly assigned as a whole to one of the preparations. In the first year, the vaccines were bivalent, containing types A (H3N2) and A (H1N1) components. In the second year, the vaccines also contained a type B component. In the first year, all viruses isolated were type A (H3N2); in the second, about three-quarters of the isolates were type B and the rest type A (H1N1). During both years, the vaccines protected the vaccinated children. Where significant differences existed, the live attenuated vaccine was more protective than the inactivated. Vaccination rates in schools in which live attenuated vaccines had been used were inversely related to illness rates of staff and unvaccinated children, suggesting that viral transmission had been reduced by the vaccine.