In the 1980s naltrexone was first studied in human alcoholics because of convincing results from animal models showing a reduction in alcohol preference when opiate receptors were blocked. The first human study was conducted by Volpicelli, et al. (1990, 1992) in male veterans in a day- treatment center. Those randomized to naltrexone showed less alcohol craving, less reported reward from alcohol, less drinking, and less relapse. This study was followed by a replication of positive results in a different population that included females and less-intense psychosocial intervention (O’Malley, et al. 1992). Since that time there have been 15 published or presented trials of naltrexone in various populations with different protocols. Thirteen of these trials showed significant benefits for patients randomized to naltrexone, but medication adherence was an issue in several of them. The protocols and the results of the published studies will be analyzed and compared. The proportion of positive naltrexone trials will be compared with the literature for antidepressant medication (Robinson and Rickels 2000). The argument will be made that naltrexone is clearly effective for a subsample of alcoholic patients, but as yet there are no reliable criteria that predict treatment response.