Self-discipline is an essential capability that supports productivity, performance consistency, and individual success in meeting increasingly complex workplace demands. However, instruments specifically developed to measure self-discipline among workers across diverse professions remain limited. This study aimed to conduct an initial validation of a Self-Discipline Instrument for workers by examining its content validity and construct validity. A quantitative approach with an instrument validation design was employed. The participants consisted of 159 workers from various professions, aged 19–46 years, selected using a convenience sampling technique. The initial instrument comprised 20 items developed based on four dimensions: impulse control, delay of gratification, perseverance and consistency of effort, and goal orientation and commitment. Content validity was assessed using Aiken’s V index, while construct validity was examined through Confirmatory Factor Analysis (CFA). The content validity results indicated that all items achieved Aiken’s V values ranging from 0.50 to 1.00, suggesting that they were suitable for further testing. The initial CFA results revealed that several indicators had outer loading values below 0.50, necessitating model refinement. The final model retained three constructs—delay of gratification, impulse control, and perseverance and consistency of effort—with a total of 11 indicators. Most model fit indices demonstrated satisfactory results (χ²/df = 1.041; RMSEA = 0.016; GFI = 0.954; AGFI = 0.926; SRMR = 0.049; TLI = 0.992; and CFI = 0.994), although the NFI value (0.868) did not meet the recommended threshold. Discriminant validity was established, with all HTMT values below 0.85. However, construct reliability and convergent validity remained suboptimal, as indicated by Cronbach’s Alpha, Composite Reliability, and Average Variance Extracted (AVE) values that fell below the recommended standards. These findings suggest that the instrument demonstrates preliminary evidence of content validity, model fit, and discriminant validity, but further refinement of the items and additional testing are required to achieve stronger psychometric properties.