The ranging accuracy of full-waveform LiDAR, also known as ranging repetition accuracy or ranging standard deviation, is affected by laser output light stability, laser pulse width, detector response time jitter, circuit noise, waveform shape, waveform sampling frequency, waveform processing algorithm and so on. The effects of different sampling frequencies and pulse widths of full-waveform LiDAR on ranging accuracy were analyzed theoretically. The waveform data under different sampling frequency (1.25, 2.5, 5 GHz) and pulse width (1, 2, 3,···, 10 ns) were collected and preprocessed by filtering, interpolation and waveform extraction. Linear Gaussian fitting, weighted linear Gaussian fitting, iterative weighted linear Gaussian fitting, expectation maximization algorithm and Levenberg Marquardt algorithm were used to calculate the ranging value and the ranging accuracy. The experimental results show that the ranging accuracy obtained by EM algorithm is least affected by waveform distortion compared with the other four algorithms, and that the ranging accuracy obtained by weighted linear Gaussian fitting algorithm is least affected by the change of sampling frequency. Under the condition of the same waveform amplitude, the actual pulse width increases 2.47 times, and the ranging accuracy obtained by EM algorithm is reduced from 0.97 mm to 1.18 mm, so increasing the pulse width will reduce the ranging accuracy. When the optical pulse width is 4 ns, the ranging accuracy of 5 GHz sampling frequency data obtained by EM algorithm is 1.71 and 3.07 times of that of 2.5 GHz and 1.25 GHz sampling frequency data, respectively, while when 2.5 GHz and 1.25 GHz data are interpolated 2 times and 4 times to 5 GHz, they are only 1.17 times and 1.29 times, so increasing the sampling frequency can improve the ranging accuracy. On the other hand, the ranging accuracy close to the high sampling frequency data can be obtained by interpolating the low sampling frequency data.