The purpose of the study was to assess medical journals' conflicts of interest in the publication of book reviews. We examined book reviews published in 1999, 2000, and 2001 (N = 1,876) in five leading medical journals: Annals of Internal Medicine, British Medical Journal (BMJ), Journal of the American Medical Association (JAMA), Lancet, and New England Journal of Medicine. The main outcome measure was journal publication of reviews of books that had been published by the journal's own publisher, that had been edited or authored by a lead editor of the journal, or that posed another conflict of interest. We also surveyed the editors-in-chief of the five journals about their policies on these conflicts of interests. During the study period, four of the five journals published 30 book reviews presenting a conflict of interest: nineteen by the BMJ, five by the Annals, four by JAMA, and two by the Lancet. These reviews represent 5.8%, 2.7%, 0.7%, and 0.7%, respectively, of all book reviews published by the journals. These four journals, respectively, published reviews of 11.9%, 25.0%, 0.9%, and 1.0% of all medical books published by the journals' publishers. Only one of the 30 book reviews included a disclosure statement addressing the conflict of interest. None of the journals had a written policy pertaining to the conflicts of interest assessed in this study, although four reported having unwritten policies. We recommend that scientific journals and associations representing journal editors develop policies on conflicts of interest pertaining to book reviews.