The aim was to test aspects of reliability and validity of a Swedish version of the Self-Efficacy for Exercise Scale (SEE-SV) for older people. Floor and ceiling effects, item-total correlation, internal consistency, test--retest reliability, content and construct validity were tested on 39 older people, mean age 72 years (SD 3.3). Content validity was also tested on five physical therapists. The results showed no floor and ceiling effects. The ICC for the total score was 0.79 (95% CI 0.61-0.89). The median weighted kappa coefficients square for the individual items was 0.63. Item-to-total correlation varied between 0.64 and 0.86 and Cronbach's Alpha for all items was >0.92. The correlations between SEE-SV and the SF-36 Mental Health and Bodily Pain subscales were r(s) = 0.50 and r(s) = 0.30 (ns), respectively. The correlation between SEE-SV and self-assessed physical activity was r(s) = 0.13 (ns). Eighty-nine percent responded that the items were relevant and 92% that no item should be excluded. All physical therapists considered the items relevant, but the written instructions could be improved. In conclusion, SEE-SV has good item-to-total correlation, satisfactory internal consistence and substantial test--retest reliability. Our results show satisfactory content validity; however, studies are needed to further evaluate construct validity and to evaluate other aspects such as predicitve validity and responsiveness.