BACKGROUND: Since physical activity and exercise levels are known to be generally low in persons with spinal cord injury (SCI), there seems to be a need for intervention. Exercise self-efficacy (ESE), the confidence persons have in their ability to be physically active and exercise, is an important and modifiable predictor of physical behavior. The goal of this study was to 1) describe ESE in persons with subacute SCI, 2) to assess ESE in subgroups based on demographic and lesion characteristics, and 3) to explore the relation between ESE and physical behavior and physical capacity. METHODS: Thirthy-seven persons with subacute SCI who are wheelchair dependent participated. Participants completed the Exercise Self-Efficacy Scale. We recorded age and lesion characteristics, measured physical behavior (physical activity, motility and sedentary day time, n = 35) with an accelerometer-based activity monitor and measured physical capacity (peak power output, n = 28 and peak oxygen uptake, n = 24) during a maximal hand-cycling test. Measurements were performed 2 months prior to discharge from inpatient rehabilitation. Mann-Whitney tests were used to test for differences between subgroups based on age and lesion characteristics and spearman correlations were used to assess the relation between ESE and physical activity and physical capacity. RESULTS: Persons with tetraplegia had lower ESE compared to persons with paraplegia (Z = -1.93, p = 0.05). No differences in ESE were found between subgroups based on age and motor completeness of the lesion. In persons with paraplegia, ESE was positively related to peak power output (rho = 0.58, p = 0.02). The relation of ESE with wheeled physical activity was rho = 0.36, p = 0.09. CONCLUSIONS: In persons with SCI who are dependent on a manual wheelchair, lesion level when categorized as paraplegic and tetraplegic affected ESE whereas age categories and completeness categories did not. Persons with tetraplegia were found to have lower confidence with regard to physical activity and exercise indicating that this subgroup can benefit from extra attention in the promotion of physical activity and exercise. In persons with paraplegia, ESE seemed to be lower in persons with less peak power output and less daily physical activity.