The effect of peak-broadening and error in interdetector volume on the local calibration curve and experimental molecular-mass averages obtained by size-exclusion chromatography (SEC) with dual concentration/viscosity detection, and determination of molecular mass using the universal calibration (UC) method, is theoretically examined using a polymer sample with a molecular-mass distribution (MMD) approximated by the log-normal function. Although peak-broadening is often neglected, its effect on the slope of the local calibration curve and, consequently, on the experimentally obtained values of the weight-to-number average ratio is large. To obtain the right values of these parameters, a numerical correction is usually recommended. While using the UC method, the relationships between the extent of peak broadening, calibration slopes and interdetector volume are complex and can contribute to the occurrence of undiscovered errors. For this reason, an understanding of this problem, using a model, is necessary. The results of the UC method are compared with those obtained using dual-detection with known Mark-Houwink-Kuhn-Sakurada parameters (MHKS method), light-scattering (LS)/concentration detection as well as with the results obtained using conventional calibration. Due to peak-broadening, the slope of a local calibration curve and the weight-to-number average ratio, (Mw/Mn)", obtained using the UC method, increase compared to the theoretical values, whereas they decrease using the MHKS or LS methods. The increase when using the UC method is even larger compared to evaluation using conventional calibration. The effect of the error in interdetector volume on the slopes of local calibrations and the weight-to-number average ratios is opposite in the UC method to that found using the MHKS and LS methods.