We report on two features of the early termination of the studies of ocular complications of AIDS monoclonal antibody cytomegalovirus retinitis trial: one related to the deliberative process of the treatment effects monitoring body (the policy and data monitoring board [PDMB]) and the other related to the relationships among the PDMB, the investigators, and the joint sponsors (the National Institutes of Health and a pharmaceutical company). The PDMB was challenged with reconciling internally inconsistent safety and mortality data and determining the weight to give mortality data from other concurrently running trials. The coordinating center faced a challenge in negotiating how to communicate results from a jointly sponsored trial. Early resolution of certain organizational and procedural issues, such as rules regarding absentee voting, to which body the monitoring committee should report officially, and the timing and general content of dissemination of different kinds of results, would make jointly sponsored trials more robust to difficulties at the final hour.