The construct validity of class ratings of instructors measured by the Course-Faculty Instrument was assessed. Two sets of variables were used which explicate aspects of the nomological network of instructor ratings: (a) three intrinsic variables of instructor behavior represented wanted sources of rating variance (e.g., perceived effectiveness of instructor communications), and (b) six extrinsic variables represented variance often considered to be undesirable, (e.g., instructor sex). Eighty-nine percent of the variance in the ratings of 129 instructors teaching 334 classes (from 10,738 student ratings) was explained by the intrinsic measures of instructor behavior. Three of the extrinsic variables correlated significantly with intrinsic variables and/or ratings (r's ≤.23). However, these latter variables (experience, rank, and publishing activities) may be as suggestive of validity as bias. © 1979, Sage Publications. All rights reserved.