OBJECTIVE In the past decade, we have witnessed the emergence of pharmacological treatments for pathological gambling with some success but many question marks. We aimed to explore pharmacological treatments that have been previously explored with some success, with the intent of comparing their efficacy and pave the way to larger placebo-controlled trials. METHODS In this study, we allocated 78 patients to 4 different types of psychotropic medications: naltrexone, topiramate, bupropion, and escitalopram. We treated patients for more than 2 years, with additional 2-year follow-ups without medication. The sample was evaluated using the 21-item Hamilton Depression Rating Scale, the Hamilton Anxiety Rating Scale, the Global Assessment of Functioning, and the Visual Analog Scale to measure general well-being before enrollment as well as at 1 month, 6 months, 24 months, and 48 months after beginning medication treatment. RESULTS During the first 2 years of treatment, 34 patients dropped out, with one more dropping out during the additional 2 years of follow-up. Significant improvement on all rating scales was seen in all groups after 2 years, except HAMD in the group that received topiramate. We found the naltrexone-treated group of patients to have a statistically significant lower dropout rate compared with other groups, statistically significant lower HAMD scores in comparison to the group treated with bupropion, statistically significant lower Hamilton Anxiety Rating Scale score compared to the groups treated with escitalopram and topiramate, and significantly higher Visual Analog Scale scores compared to the groups treated with bupropion and topiramate. CONCLUSIONS Pathological gambling is essentially a biopsychological disorder that may be attenuated provided that patients adhere to medication. In our study, among 4 medications with different mechanisms of action, naltrexone was found to be the most effective. Placebo-controlled studies involving large numbers of subjects are required before naltrexone's efficacy can be established.