In this study, we investigated job stressor ratings through Lazarus' transactional stress theory and the usefulness of supervisor ratings as an alternative to employee self-reports. Based on the finding that negative affectivity (NA) causes incumbents to over-report job stressors, we hypothesized that supervisors may also be affected by their NA when appraising an incumbent's job stressors. Building upon the literatures on judgement processes and social-cognitive information processing, we further hypothesized that stressor observability is an important boundary condition. Specifically, we hypothesized that the impact of NA on both self- and supervisor-reported job stressors should increase as stressor observability decreases. Moreover, we hypothesized that incumbent and supervisor ratings would converge less as stressor observability decreases. Data from 260 incumbent-supervisor dyads showed that stressor observability reduced the impact of supervisor NA on supervisor ratings (but not the impact of incumbent NA on incumbent stressor ratings). Further, as hypothesized, incumbent and supervisor ratings showed less convergence the less observable the stressor was. The results highlight the importance of personal and situational factors in the stressor appraisal process – an issue that advances both research and practice in the field of stressor ratings.