The molecular masses of three rat-plasma carboxylesterases (ES-1, ES-2, and ES-14) were estimated by transverse-gradient polyacrylamide gel electrophoresis and subsequent application of Ferguson-plot-based calculation methods. Two electrophoretic buffer systems were used and the data subjected to either weighted or unweighted regression analysis. The Tris-boric acid buffer system produced significantly higher retardation coefficients than the Tris-glycine system. Molecular mass estimates were significantly higher with the Tris-glycine buffer system. Unweighted instead of weighted analysis produced significantly higher molecular mass estimates. Molecular mass estimates also depended on the calculation method, that is, the choice of calibration relationship with molecular size as a function of retardation coefficient. Three commonly used calibration relationships were compared. On the basis of their accuracy, both the weighted log[retardation coefficient] versus log[molecular mass] plot and the square root of retardation coefficient versus molecular radius were found suitable, provided that the Tris-boric acid buffer was used for electrophoresis. Using the former calibration relationship, the molecular masses of rat-plasma ES-1, ES-2, and ES-14 were 55.5, 61.1, and 65.3 kDa, respectively.