Abstract Background Although several studies have examined the association between occupational stress and turnover, these studies relied on cross-sectional designs, subjects’ self-report, healthcare workforce, or small sample sizes. This study aimed to confirm whether occupational stress increases the risk of turnover in a large-scale prospective cohort study using actual turnover data from company records. Methods The participants were 3892 male and 5765 female employees aged 20–49 years in a financial service company. We followed them from October 2012 until April 1, 2016 and used company records to identify employees who resigned. We identified employees with high and low stress using the Brief Job Stress Questionnaire. Hazard ratios for turnover in high-stress employees were estimated using Cox proportional hazards models, and population attributable risks were calculated separately for men and women. Results During 11,475,862 person-days, 122 men and 760 women resigned. After adjustment for age, length of service, job type, and position, the hazard ratios (95% confidence intervals) for turnover in high-stress employees were 2.86 (1.74–4.68) for men and 1.52 (1.29–1.78) for women. The corresponding population attributable risks for high stress were 8.2% for men and 8.3% for women. The component scores, i.e., job stressors, psychological/physical stress response, workplace social support, and job strain (the combination of high job demands and low job control) were also significantly associated with turnover (p < 0.05). Conclusions Occupational stress increases the risk of actual turnover. Measures to prevent occupational stress may be useful to prevent employee turnover.