In this article the relationship of the hospital psychiatrist and the agencies requesting information and/or recommendations concerning patients has been examined. Two questions immediately arose: 1. Whose agent is the psychiatrist? and 2. Is he oriented to therapy or to public service or does he believe that he can encompass both objectives?The types of requests for information and recommendations were enumerated. It has apparently been assumed that many requests are reasonable and should be answered. The problem then was to determine which ones were reasonable and to decide how they should be answered. It was suggested that the first issue should be that of questioning whether the psychiatrist should supply any information. This was then considered in terms of its effect on psychotherapy. Obviously, therapy will be altered if the patient sees the psychiatrist as a possible informer as well as a helper.If information is supplied, is it really useful? This brings us face to face with the issue of how well predictions can be made in instances involving many variables, some known but many unknown, which can be arranged in an exceedingly large number of combinations. Some comments were, also, made on the misleading effect of labelling.The expectation that requests would be answered and the practice of complying, in part, stems from the general practice of medicine. Too little attention has been paid to significant social factors. It was suggested that psychiatrists might be seduced to claim that they possessed special ability to foretell the future. As a result "educated" guesses might be dispensed as facts.During a period of hospitalization, the patient's family might have to be provided with information to participate in immediate decision-making. This stands in sharp contrast, however, to making decisions about the patient's ability to work or attend school after he has left the hospital and is assuming responsibility for his own welfare.