After an outbreak of acute rheumatic fever at a US Army training installation, a benzathine penicillin G prophylaxis program was instituted. Surveillance data were analyzed to measure rates of febrile, acute respiratory disease (ARD) among trainees before and after prophylaxis was begun. Annual admissions for ARD decreased from 1927 to 690 (-64.2%) after benzathine penicillin G prophylaxis was begun. Admissions with throat cultures positive for Streptococcus pyogenes fell from 595 to 63 (-89.4%), a reduction that accounted only for a minority (43%) of the total 1237 "prevented" admissions. Temporal changes in disease rates at other installations where drug was not administered were also analyzed. Only a small decrease in the number of annual ARD admissions (-6.3%) was observed at other training installations. These findings support a hypothesis that benzathine penicillin G has a broad effect in the prevention of ARD that extends beyond the simple elimination of group A streptococcal infection.