In contrast to the widely used assessment approach in which self-esteem stability is measured as the standard deviation of repeated measurements, direct measurements of self-esteem stability have hardly ever been implemented in longitudinal studies. The primary goal of the present study was to examine the temporal stability and predictive validity of a direct assessment of self-esteem stability compared with the trait level of self-esteem with respect to the prediction of psychological distress (PD). We examined a sample of 136 employees who completed self-report measures of both self-esteem level [Rosenberg Self-Esteem Scale (RSES)] and self-esteem stability [Self-Esteem Stability Scale (SESS)] along with a measure of PD (SCL-90) at two time points across an interval of 1 year. The results underline the relevance of perceived self-esteem stability in the temporal prediction of PD: After controlling for initial PD, we found that self-esteem stability predicted PD better than self-esteem level did. Therefore, we recommend that the RSES be expanded by adding the three SESS items that directly measure the stability of self-esteem.