A new approach to polychotomous scoring of test items, similar to "max-alpha" scaling (MAS) and known as polyweighting, has been developed. Unlike MAS, this new method of polychotomous scoring provides scoring weights for a given item that are independent of the difficulty of other items in the analysis. Moreover, the scoring weights are bounded so that an examinee can never receive more credit for an incorrect response than for a correct response. Polyweighting assigns a score weight to each item response category that is approximately equal to the mean percentile rank of examinees who chose that category in the item calibration sample. An iterative procedure is used to derive polyweights for a set of items. In a computer program written for this process, iterations continue until the mean squared correlation ratio between items and percentile ranks stops increasing. To evaluate the effect of polyweighting on the reliability and domain validity of test scores, data from 1,100 resident physicians completing a 200-item test on otolaryngology were obtained. Results indicate that polyweighting can provide a higher level of test reliability and domain-related validity than can traditional number/proportion-correct scoring. Further research should compare the effectiveness of polyweighting with item response theory methods. Seven tables and seven graphs are presented. (TJH)