Observational ratings of student clinical performance are influenced by factors other than the quality of the performance. Individual raters may be more stringent or lenient than their colleagues. In this medical school setting, multiple raters evaluated each student. To reduce the influence of "error" due to differences among raters, each rater was assigned a handicap score which was calculated in three steps: (1) identify the cohort of students observed by the rater, (2) calculate the mean of all faculty ratings for that cohort (grand mean) and the mean given those students by the rater, and (3) subtract the individual rater mean from the grand mean. Analysis of the "original" and "adjusted" ratings for two academic years indicated no differences in overall mean and standard deviation. Generalizability analysis indicated an improvement equivalent to increasing the number of raters per student by 50 percent (i.e., the variance component due to error was reduced by about 33 percent). (Author)