Thirteen patients with chronic osteomyelitis, treated for 6 months with rifampin, had serial 99mTc phosphate and 67Ga scans to determine their value in assessing response to treatment. In patients who responded to treatment, gallium scans were deemed more accurate than 99mTc phosphate bone scans. The gallium scans, although still abnormal at the end of 6 months of antibiotic therapy, showed an improvement trend in all the responders except one in whom fracture recurred. Worsening or lack of improvement on gallium scans predicted active bone infection in five of six "clinical-failure" patients who had documented active bone infection. 67Ga scans eventually became normal in all patients who remained asymptomatic (excluding one with recurrent fracture). 99mTc phosphate scans became normal in only one of five clinical responders. All nonresponders had persistently abnormal scans, although after 6 months of therapy only four of seven showed worsening or no improvement on the scan. Therefore, 67Ga is preferred over 99mTc phosphate bone scans in the assessment of response to therapy in chronic bone infection. Clinical utility of the gallium scan is most significant in patients whose clinical assessment is uncertain, but routine use of this technique does not appear to be warranted. Gallium images are most valuable when obtained over a period of time, so that the trend of improvement versus nonimprovement is evident.