Drowsiness of vehicle operators is a major hazard in transportation systems, and methods need to be developed for practical evaluation of drowsiness level. One suggested approach is observer rating. Accordingly, an experiment was carried out using trained observer-raters to evaluate the levels of drowsiness of drivers, the drivers' faces were recorded on videotape. Videotaped segments of drivers at various stages of drowsiness were presented in two sessions separated by a time interval of one week. The experiment was directed at determining test-retest reliability, interrater reliability, intrarater reliability, and consistency. Results indicate that such ratings are reliable and consistent. A subsequent experiment shows that ratings covary with other known indicators of drowsiness.