OBJECTIVE: Effectiveness trials are designed to evaluate patients in their natural setting with fewer constraints than in efficacy trials. The less-structured environment can result in a failed trial if participant activity is unknown. We prepared for this possibility by including electronic monitoring of medication dosing in a multi-center trial. METHODS: The trial was designed to assess the effectiveness of naltrexone for the treatment of chronic alcoholism. Patients took either naltrexone or placebo once daily, using MEMS caps (APREX, Union City, CA) on their medication bottles to record the date and time of each opening. We planned analyses by intention-to-treat and covarying compliance as continuous and categorical variables (grouped as taking medication during 0-24%, 25-49%, 50-74%, >75% of weeks). RESULTS: Primary endpoints showed no differences between treatment groups at 3 months. Electronic monitoring revealed that patients took 72+31% of naltrexone and 70+31% of placebo doses (overall compliance rates). Naltrexone was taken by 13%, 11%, 12%, and 65% of patients by category. Placebo was taken by 14%, 14%, 11%, and 61% of patients by category. Compliance rates were not significantly different overall or by category between treatment groups. Planned secondary analyses demonstrated that compliance was a predictor of success (p=0.03 for drinks/day), with no interaction for treatment. CONCLUSION: These data demonstrate the value of electronic compliance measurement that provided data on any period needed for analyses. Without these data, the results of a complex and expensive study would have been questioned. Critics could have charged that compliance rates differed among treatment groups, or that inadequate amounts of medication were taken to assess outcomes.