Bacteremia was documented in 19 (76%) of 24 patients with sepsis caused solely by decubitus ulcers, persisted in all but two, and was polymicrobial in 10 (42%). Obligate anaerobes were isolated from 12 patients (63%) and included Bacteroides fragilis in 11 (58%). Aerobes, primarily Proteus (21%) and Staphylococcus (16%), were isolated in nine patients (47%). Therapy was judged appropriate when the microbial isolates were susceptible in vitro to the antibiotic used. Nineteen patients received clindamycin plus gentamicin, which was considered appropriate for all but one patient. Four patients received cephalothin plus kanamycin, which was inappropriate for three patients. One patient received appropriate treatment with methicillin, gentamicin, and chloramphenicol. Patients who received appropriate antibiotics and had surgical intervention had the lowest mortality rate (14%); in those treated with appropriate antibiotics but without surgical intervention, the incidence of death was 67% (P less than 0.05). Patients who received inappropriate antibiotics, whether or not there was surgical intervention, had a 75% mortality rate (P less than 0.05). Surgical debridement and antibiotic therapy effective against aerobic as well as anaerobic bacteria are important factors in the treatment of sepsis caused by decubitus ulcers.