An improved statistic (weighted Kappa) was tested for reviewing the reliability of the palpation of pedal pulses by three observers. Significant improvement was noted after training. Reliability also persisted with respect to a reduced form of the conventional ordinal grading system. The findings indicated that problems of usefulness or reliability may arise when more than five distinctions are made in describing the pulses or when the data are completely reduced to only a dichotomous (present-absent) system.