I describe an approach to predicting competence in technical skills for the purposes of resident selection. To demonstrate a predictive relationship, it is necessary to use measures that exhibit variation, reliability, and validity. There is little evidence that such measures are routinely used in the process of selecting residents. I argue that the selection of assessment instruments in the predictor domain must be guided by relevant theoretical considerations, while assessment in the surgical domain must make use of more objective and reliable instruments than is currently the practice. I present a brief summary of research on predicting operative technical competence.