An intervention designed to enhance preaction self-efficacy beliefs (i.e., beliefs about ability to initiate behavior despite anticipated barriers during the initiation period) was tested in patients with spondylosis in relation to initiation of exercises recommended by a consultant in orthopedic rehabilitation. Sixty patients (age 28–83 years; 44% men) with spondylosis who had not previously performed exercises recommended for degenerative spine diseases were randomly assigned to a control (education session) or intervention group. Three weeks later, intervention patients performed recommended exercises more frequently than controls. Regression analysis for all patients showed that preintervention, preaction self-efficacy predicted exercise. Age and preintervention self-efficacy moderated the intervention effects. Among older patients, only those with weak preintervention, preaction self-efficacy beliefs benefited from the intervention, whereas among younger patients, only those with strong preintervention, preaction self-efficacy beliefs benefited from the intervention.