A neurobiological model of risk-taking suggests that differential timing in the maturation of the brain networks associated with emotional processing and cognitive control may predispose adolescents to risky behavior, including alcohol and other substance use. Heavy alcohol use during adolescence has been shown to alter normative brain functioning, though it remains unknown whether alterations normalize with sustained abstinence or persist after cessation of use. The present study utilized fMRI to examine the effects of heavy alcohol use and short-term abstinence on adolescent neural functioning during a risky decision-making task. Heavy drinking adolescents and non-users completed three neuroimaging assessments, spaced two weeks apart (baseline, +2weeks, +4weeks). Adolescents abstained from alcohol and other substances for the duration of the study, confirmed through regular urinalysis screenings. During scanning, participants completed a modified Balloon Analog Risk Task (BART) to inflate balloons by entering a fixed number of "pumps". Adolescents earned 1cent/pump unless the balloon popped according to a predetermined value; a higher pump number represented a riskier choice. Relevant neuroanatomical regions of interest were identified for each phase of decisionmaking (assessment, anticipation, and evaluation of outcome) and between-group differences in blood oxygenated level dependent (BOLD) response were assessed at baseline. In addition, longitudinal analyses examined the main effects and interaction of Group and Time on BOLD response across the five-week period of abstinence. At baseline, heavy drinkers showed less BOLD response in the right insula during anticipation [t(1,39) = -2.89, cluster = 324 _l, p<.01] and more BOLD response in the ventromedial prefrontal cortex [VMPFC; t(1,39) cluster 1 = 3.17, 702 _l; t(1,39) cluster 2 = 3.24, 405 _l, ps<.01] during evaluation of negative outcomes, compared to non-users. These differences were no longer evident at either follow-up time point. However, significant main effects of Group and interaction effects (Group X Time) were observed in other regions. Averaged across time, heavy drinkers showed reduced BOLD response in the dorsolateral prefrontal cortex (DLPFC) during the assessment phase [14 clusters, F-statistics (1,40) ranged from 7.65 to 9.85, volumes ranged from 378 _l to 1,782 _l, ps<.01] and the left insula during anticipation [F(1,40) = 8.25, cluster = 297 _l, p<.01], and greater BOLD response in the left VMPFC during anticipation [F(1,40) = 10.94, cluster = 756 _l, p<.01]., compared to non-users. For non-users, BOLD response in the right anterior cingulate increased across time during the assessment phase [t(1,15) = -3.11 when previous balloons popped; t(1,15) = -3.48 when previous balloons did not pop, ps<.01], while for heavy drinkers, BOLD response increased across time in the right VMPFC/anterior cingulate during anticipation [t(1,19) = -3.33 when previous balloons popped; t(1,19) = -3.54 when previous balloons did not pop, ps <.01]. Overall, these findings highlight differential neural functioning during risky decision-making in heavy drinking adolescents and non-users. While group differences in BOLD response observed at baseline were no longer apparent after two weeks of abstinence, other differences persisted across a five-week period of sustained abstinence. This pattern of results suggests that alterations in neural functioning commonly observed in adolescent alcohol users may result from a combination of acute changes related to use as well as pre-existing vulnerabilities. Conversely, some brain functioning abnormalities may reverse after longer periods of abstinence.