One hundred fifty-three problem drinkers were randomly assigned to receive naltrexone 50 mg or placebo on a daily or targeted (to high-risk drinking situations) basis. Using structured nightly diaries, participants recorded negative and positive mood, desire to drink, and alcohol consumption over 8 weeks. Results indicated that individuals engaged in any drinking and heavy drinking more on days characterized by relatively higher levels of positive or negative mood states. Naltrexone attenuated the positive association between heavy drinking and both positive and negative mood, and targeted administration attenuated the positive association between heavy drinking and positive mood. There was also evidence that desire to drink mediated the effect of targeted administration on the relation between positive mood and any drinking that day. These findings underscore the utility of daily measurement for understanding the processes that underlie pharmacological interventions for problem drinking.