Often, more than twenty generic (GE) drugs are developed for one brand name (B-N) drug. Each GE drug has passed the trial of bio-equivalence to the B-N drug, and its characteristics (which are generally expressed by AUC, Cmax, and Tmax, or one of its subsets) are close to the values for the B-N drug in each trial. However, trial-to-trial variation in the AUC, Cmax, and Tmax values is substantial. A model is developed in this paper to extract the distance between the characteristics of the GE and B-N drugs by adjusting for trial-to-trial variation. A method is proposed in this paper to use this distance for selecting better GE drugs from the list of candidates.