In this evaluation eight observers viewed small topographical defects in paint samples out of doors and in a mock inspection facility. Each observer rated defect severity and the probability of returning a new vehicle with these defects to the dealer for repair. Severity ratings for defects viewed under the recommended inspection lighting were the same as those made outdoors. A nonlinear model relating the probability of a repair request to defect severity (R2 = .95) is provided. A method of measuring the essential physical attributes of the defect, specular highlight area and modulation, is described. A second model relating defect severity to the essential defect attributes (R2 = .88) is provided.