The challenge–hindrance stressor framework can be used to examine learning in the workplace. However, two issues exist regarding the association between challenge stressors and learning: (1) there are two hypotheses related to challenge stressors and learning: a linear association hypothesis based on action regulation theory and an inverted U-shaped hypothesis based on the comfort zone model, and (2) few studies have examined the association with learning using longitudinal data. This study examined the within-person effects of a challenge stressor on learning and strain using Japanese panel data obtained annually for eight years. The findings reveal that cognitive demands have a linear relationship with learning and that workload has an inverted U-shaped relationship with learning. Our findings suggest that within-person effects may be more likely to benefit from challenge stressors than between-person effects. Moreover, depending on the type of job demands, the assumptions of the challenge-hindrance stressor framework should be improved, and the assumption that there is a linear relationship between learning and workload may need to be revised.