OBJECTIVES: The physical activity paradox suggests that occupational physical activity (OPA), unlike leisure-time physical activity (LTPA), may detrimentally impact health. We explored the relationships of OPA and LTPA with work ability (WA) and health-related productivity loss (HRPL). METHODS: This study included 5,501 workers in Korea who were recruited in 2021 through a web-based cross-sectional questionnaire. The questionnaire was utilized to quantify OPA and LTPA in metabolic equivalents, while WA and HRPL were also measured. Non-parametric regression, using a generalized additive model (GAM), was employed to visualize the relationships of LTPA and OPA with WA and HRPL. Mean differences in WA and HRPL, in relation to OPA and LTPA, were examined using linear regression models. These models were adjusted for covariates including sex, age, body mass index, education level, alcohol consumption, smoking history, insomnia, occupation, hours worked, and income. RESULTS: The GAM and linear regression analyses revealed that higher LTPA corresponded with higher WA and lower HRPL. In contrast, as OPA increased, WA decreased and HRPL increased. However, within the group with high OPA, HRPL was not significantly lower in the high-LTPA subgroup relative to the low-LTPA subgroup (mean difference=1.92%, p=0.343). This pattern was especially pronounced among workers aged 60 years and older, with an increase in HRPL observed with increasing LTPA among the respondents with high OPA. CONCLUSIONS High LTPA levels were associated with elevated WA and diminished HRPL. In contrast, higher levels of OPA were associated with lower WA and higher HRPL.