The purpose of this article was to describe the protocol by which predictive instruments can be tested for validity and to evaluate the usefulness of an instrument for predicting pressure sore risk in an AICU. The Braden Scale for Predicting Pressure Sore Risk was described. Methods for measuring predictive validity and for calculating sensitivity, specificity, and per cent predictive value of positive and negative tests were discussed. Sixty consecutively admitted AICU patients who were pressure sore free were rated for pressure sore risk within 24 to 72 hours after admission. The skin condition of each patient was systematically assessed every 2 days. Twenty-four subjects developed pressure sores during the study period. The critical cut-off point at which the patient could be judged to be at risk for pressure sore formation was a Braden Scale score equal to or less than 16. The sensitivity and specificity of the scale at this score were 83 to 64 per cent, respectively. The per cent predictive value of a positive and negative test were 61 and 85 per cent, respectively. The Braden Scale compared favorably with the Norton Scale in respect to sensitivity. The specificity, or the tendency of a scale to overpredict, was greater for the Norton than for the Braden Scale. The Norton Scale overpredicted by 64 per cent, whereas the Braden Scale overpredicted by 36 per cent. This difference may be important clinically if all patients who were judged to be at risk received additional nursing care or protective devices. A greater number of patients may receive unnecessary and expensive treatments using the Norton Scale.