The effects of variables on treatment of the gel permeation chromatogram are reported. Variables investigated include (1) the molecular weight distribution of polymers for preparing the calibration curve, i.e., the logarithm molecular weight-elution count relationship, (2) nonlinearity of the calibration curve, and (3) fluctuation of the baseline. The deviation of the calibration curve prepared by the polymer having broad molecular weight distribution was evaluated in detail by assuming log-normal distribution function for the distribution. The polymer having a D value less than 1.3 was recommended for this purpose. Generally, the shape of the chromatogram is fairly different from that of the true molecular weight distribution curve when the calibration curve is not linear over the entire range of interest. By fitting polynomials to the calibration curve, the chromatogram was sufficiently converted to the molecular weight distribution curve. The apparent difference between them was removed. Slight deviation of the baseline from the true one gave rise to obvious error in the calculated molecular weight and its distribution, especially for the sample having a broad distribution.