BACKGROUND: Evidence-based physical activity interventions that can be delivered to large numbers of adults at an acceptable cost are a public health priority; website-delivered programs have this potential. The purpose of this study was to systematically review the research findings and outcomes of website-delivered physical activity interventions and to identify relationships of intervention attributes with behavioral outcomes. METHODS: A structured search of PubMed, Medline, PsycInfo, and Web of Science was conducted for intervention studies published up to July 2006. Studies included in the review were those that (1) used websites or e-mail, (2) had physical activity behavior as an outcome measure, (3) had randomized controlled or quasi-experimental designs, (4) targeted adults, and (5) were published in English. RESULTS: Of the fifteen studies reviewed, improvement in physical activity was reported in eight. Better outcomes were identified when interventions had more than five contacts with participants and when the time to follow-up was short (</=3 months; 60% positive outcomes), compared to medium-term (3-6 months, 50%) and long-term (>6 months, 40%) follow-up. There were no clear associations of outcomes with other intervention attributes. CONCLUSIONS: A little over half of the controlled trials of website-delivered physical activity interventions have reported positive behavioral outcomes. However, intervention effects were short lived, and there was limited evidence of maintenance of physical activity changes. Research is needed to identify elements that can improve behavioral outcomes, the maintenance of change and the engagement and retention of participants; larger and more representative study samples are also needed.