Background <p>Japan lacks efficient assessment tools for occupational balance (OB) in people with mental illness (MI).</p> Aims/objectives <p>We investigated the test–retest reliability and construct validity of a Japanese version of the Satisfaction with Daily Occupations and Occupational Balance (SDO-OB-J) for people with MI.</p> Methods <p>SDO-OB-J interviews were conducted twice at a 2-week interval.</p> Results <p>Fifty-three participants joined. Regarding test–retest reliability, the weighted kappa coefficients were 0.75 for total activity level, 0.85 for total activity satisfaction, and 0.54 for general OB. Concerning construct validity, activity level and activity satisfaction had no significant correlation with global functioning, neurocognition, or mental-related quality of life (QOL). Physical-related QOL had a significant correlation only with activity satisfaction. General OB responses showed no neurocognitive differences, and the in-balance group had higher mental-related QOL than the under-occupied group.</p> Conclusions <p>Weighted kappa coefficients were classified as substantial for activity level, almost perfect for activity satisfaction, and moderate for general OB. Activity level demonstrated discriminant validity with global functioning, neurocognition, and QOL. Activity satisfaction displayed discriminant validity with global functioning and neurocognition and convergent validity with physical-related QOL. General OB showed construct validity. The SDO-OB-J showed good test–retest reliability and construct validity in people with MI.</p> Study registration ID <p>UMIN000050730</p>