Although knowledge about relapse prevention is still at an early stage, the extant data highlight the importance of several constructs. 1. Motivation for abstinence remains central. The construct itself is often clouded because of its association with mystical notions such as willpower and self-control. We know that manipulation of environmental events can increase motivation. These interventions are effective, however, only as long as the contingencies are in effect. We need to develop and evaluate strategies for transferring contingency management to the natural environment, that is, to institutions and groups that can perpetuate them for the long term. Also, clarification of the kinds of abstinence goals needed to prevent relapse is important. 2. Coping skills have been studied by several investigators, but research on these, except for job-finding skills, is not encouraging. The skills usually taught may be too basic. Skills training oriented to complex targets, such as building nondrug-using networks, may be useful and should be further explored. 3. Social support is clearly important, yet we do not know how best to use it to promote abstinence. The little research available suggests that both familial and nonfamilial systems should be mobilized. We need to define abstinence-promoting supportive behaviors, identify and engage important support systems in treatment, and help patients expand their nondrug-using contacts. 4. Negative affect may be causally related to relapse. We need to continue efforts to identify dysphoric patients and develop interventions to ameliorate dysphoria concurrent with drug abuse treatment (cf. Zweben and Smith 1989). 5. Drug cue reactivity and extinction to drug cues have been demonstrated in the laboratory. What is needed in this promising line of research are (1) investigation of cues and cue-reactivity phenomena in the natural environment or in conditions closely mimicking that environment and (2) extinction methods that transfer from the treatment setting to the outside world. Other phenomena are not well understood but seem intuitively important. Maladaptive ways of responding to lapses, such as the AVE, are included here. Another is stress, which our patients and our clinical intuition tell us must play a role in relapse. Its exact role is far from clear.