We have examined the validity and reliability of Japanese SRS-24 and SF-36 in patients with scoliosis. 278 patients with scoliosis were recruited. Their average age was 24.8 years (range 17-84 years) and 83% of patients had idiopathic curves. The major coronal curve averaged 43 degrees (range 10 degrees -114 degrees). Questionnaires surveyed were SRS-24 and SRS-22, and SF-36. Only domains applicable to untreated patients were analyzed. Scale level analysis was evaluated by the ceiling and floor effect. The Cronbach alpha and item-scale correlations were calculated as representatives of internal consistency reliability and item internal consistency, respectively. Most domains of SF-36 had large ceiling effects. Both the SRS questionnaires had limited ceiling/floor effects. SRS-22 had a better internal consistency than SRS-24. Low item-scale correlations (<0.25) were found in question 14, 15, and 18 of SRS-22. The Japanese version of SRS-22 was the most useful measure for patients with scoliosis. However, some questions have low item-scale correlations, lowering reliability of the Function domain.