To measure the phase retardance and the fast-axis position angel of non-achromatic waveplates accurately, a high-precision measurement system is established based on the fitting light intensity method and spectral analysis method, which can realize the high-precision measurement of the retardance of non-achromatic waveplates at 0°~360°. The measuring system is composed of a white light source, an aperture diaphragm, a narrow band filter, a Glan Taylor prism as polarization generator, a non-achromatic waveplate to be measured, a Glan Taylor prism as polarization analyzer, an optical power meter or spectrometer. When the phase retardance and the fast-axis position angel of non-achromatic waveplates are measured by the fitting light intensity method, the white light emitted by the white light source is collimated parallel by the collimating system, then it passes through an aperture diaphragm, a narrow filter and a polarization generator, which is modulated into monochromatic linearly polarized light. Next the linearly polarized light goes through the waveplates and the polarization analyzer, is finally received by the optical power meter. During the whole measuring process, the high precision motor drives non-achromatic waveplates that are to be measured to rotate uniformly. However, in the measurement process of spectral analysis measuring method, it is necessary to move the narrow filter out of the optical measurement path, and replace the power meter with the spectrometer as the terminal detection device, in this case, non-achromatic waveplates are not needed to rotate by the motor. Before the formal measurement of the retardance and the initial fast-axis position angle of waveplates, the stability of the measurement system and the sources of measurement errors are analyzed in detail in this paper. Under the light intensity measurement method, the influence of random errors such as light source intensity jetter, the initial fast-axis position angle and the rotating position angle of the rotating waveplate and some system errors like nonlinear effects of photoelectric response of the powermeter and deviation of collimated beam is simulated. And the reason why the fitting light intensity method can not measure the retardance at 180° of non-achromatic waveplates accurately is also analyzed detailly. Under the simulation of the spectral analysis method measuring the retardance of waveplates, we also simulate the influence caused by some random errors such as light source intensity jetter, wavelength monochromaticity and motor positioning errors. Then we suppressed the random errors mentioned above in the two measurement methods in the laboratory. And the nonlinear effect of the photoelectric response of the detector is also corrected. Finally, we measured the non-achromatic $\lambda /\mathrm{4}$waveplate, $\mathrm{0.356}\lambda$ non-achromatic waveplate and $\lambda /\mathrm{2}$ non-achromatic waveplate at 632.8 nm by using this measurement system, and the measuring data were fitted by the nonlinear least squares method. The results of the phase retardance and fast-axis position angle of the non-achromatic waveplates were obtained. It can be seen from the actual measurement that the measurement accuracy of the retardance of the $\lambda /\mathrm{4}$ waveplate and $\mathrm{0.356}\lambda$ waveplate by the fitting light intensity method is relatively high, and the measurement error is less than 0.05°. The measurement accuracy is more than one order of magnitude higher than the traditional light intensity measuring method. Moreover, the measuring device of the fitting light intensity method is simple,fast and easy to operate. For the non-achromatic $\lambda /\mathrm{2}$ waveplate, the spectral analysis method is adopted to measure the retardance under this system. in this case, the terminal device performed by the powermeter in the fitting light intensity method is switched by the spectrometer and Charge Coupled Device camera. And the narrow band filter is also removed in the light path. From the measurement results, we can know that the measuring error of the retardance is less than 0.02° by this measuring method, which is much smaller than the 0.70° of measuring error caused by the fitting light intensity method. So the spectral analysis method overcomes the defect that the retardance at 180° of the non-achromatic $\lambda /\mathrm{2}$ waveplate can not be accurately measured by the light intensity method. Therefore, the phase retardance of non-achromatic waveplates from 0°to 360° and the fast-axis position angel can be measured accurately by the measurement system designed in this paper, which provides the basis for the accurate measurement of the polarimeter.