This study examined the interrater reliability and validity of a newly developed test of physical work abilities, the Physical Work Performance Evaluation. Eleven physical therapists were trained to administer and score this evaluation. From this group, two therapists at a time simultaneously and independently evaluated 50 patients with musculoskeletal disorders as they performed the tasks of the Physical Work Performance Evaluation. At the conclusion of the evaluation, each therapist determined the safe level of physical work for each patient. A comparison of the two independent evaluations was used to determine reliability. To determine validity, the predicted level of work was compared with the actual level of work. Kappa coefficient between the two therapists on the level of work was .83. Spearman rho correlations between the predicted and actual levels of work ranged from .41 to .55. Only 14 to 18% were working above the level predicted by the Physical Work Performance Evaluation. These results indicate high interrater reliability. Given the lack of a perfect standard for validity comparisons, these results also provide evidence in support of convergent validity. The test can be used in making decisions regarding return to work after injury, preemployment placement, and vocational exploration.