The transtheoretical model of behavior change (TTM) is often used to understand changes in health-related behavior, like exercise. Exercise behavior in fitness clubs is an understudied topic, but preliminary studies showed low frequencies and large numbers of drop-out. An initial 12-week self-efficacy intervention reported significant effects on exercise behavior. The objective of this follow up study is testing effects on exercise behavior over 52 weeks and the long-term relationships of all TTM constructs. In total 122 participants (M_age 42.02 yr.; SD 12.29; 67% females) were recruited and randomly assigned to group 1 (control), group 2 (self-set activities) and group 3 (self-set goals coaching). All participants were monitored 52-weeks. Measurements at baseline, 4, 8, 12, 26 and 52 weeks, using validated scales for stages of change, self-efficacy, decisional balance and processes of change. Exercise behavior and drop-outs were registered. An ANOVA revealed that group 3 significantly (p < 0.05) differed in exercise sessions from group 1 and 2 during the 12 weeks. A chi-square test indicated significant differences for continuing exercising after the intervention: 7 of group 1; 6 of group 2; 19 of group 3. In total 5 demonstrated regular exercise behavior at 26 weeks, and 3 at 52 weeks. Self-efficacy, decisional balance, and processes of change showed limited long-term changes over the later stages of change. At all measurements, participants reported more pros than cons and used more behavioral than cognitive processes. Exercise behavior of members in fitness clubs demonstrated dramatic developments in 52 weeks. The frequencies of sessions were so low that health effects will be minimal. The integrative character of the TTM appears to be weak; the data indicated limited relationships. More research is needed to understand exercise behavior and define optimal strategies to increase exercise attendance and decrease drop-outs in the long term.