This study evaluates compliance with taking oral cotrimoxazole in an ALRI control programme in rural Bangladesh. Health workers administered the first dose to children with moderate disease and entrusted relatives to give the remaining doses. A team of medical assistants visited the families of cases 3 to 5 days after initiating treatment and counted the remaining tablets. Medical assistants undertook 367 visits to families of children under treatment at a mean (SD) of 4.4 (0.99) days after treatment began. All children appeared to have been given the antibiotic, but one-quarter were being under-dosed on the day of the visit. Under-dosing did not correlate with any of the socio-demographic variables studied, and seemed to be homogeneously distributed in the community. Under-dosed children did not seem to have a higher risk of subsequent ALRI episodes during the study period. There was no indication of progression to severe disease or death in home-managed cases of moderate pneumonia in this study. These findings raise a question about the need for 5 days of oral antibiotic in the management of moderate pneumonia. Careful studies of the effect on subsequent morbidity, mortality and antibiotic resistance of providing briefer treatment for moderate episodes of ALRI are required. If a briefer course proves effective, this would have important implications for funding programmes of control of ALRI in the community.