There are approximately 13 million cancer survivors in the U.S.; fewer than 10% meet the exercise recommendations outlined by the American College of Sports Medicine. Poor adherence is likely explained by a variety of factors. If we can identify these factors, we may be able to tailor the prescription, monitoring, and encouragement of exercise more effectively. PURPOSE: To determine variables that influence retention in an exercise trial on cancer survivors. METHODS: We enrolled 157 cancer survivors in an exercise program lasting 10 weeks and consisting of biweekly cardiovascular, strengthening, and flexibility components. At baseline and following the intervention, we assessed anthropometric and cardiovascular profiles, health and cancer history, and physical functioning. Chi-square and logistic regression analyses tested variables associated with program completion. RESULTS: We retained 37.7% of patients through follow-up. Women were more likely to complete the trial (43.2%) than men (19.4%; p=0.010). Differences between cancer type were minimal. Adherence was better among breast cancer patients (p=0.016) but this was attributable to sex; there was no difference among patients with multiple cancers (p=0.583) or patients who had a previous heart attack (p=0.681) or stroke (p=0.528), had diagnosed hypertension (p=0.513) or pulmonary disease (p=0.199), were obese (p=0.893), or smoked (p=0.333). Fatigue (p=0.696) and mode of treatment (surgery, chemotherapy, radiation; p>0.225) did not affect completion. There was a difference among patients with hyperlipidemia (50.0% retained) compared to patients without hyperlipidemia (32.4%; p=0.040). Patients with poor sit-and-reach scores were also more likely to drop out: 53.3% of patients who could reach their toes completed the program compared to 26.5% who could not reach their toes (p=0.016). CONCLUSION: Exercise adherence is low among cancer survivors; in our sample, fewer than 40% of patients were retained through follow-up. Several factors predicted retention, but sex had the strongest association. Further efforts must be made to identify risk factors for attrition in this population. The differences observed in retention by sex suggest other cohorts may need to be stratified by sex to verify our findings.