In homogeneous atmospheres, backscatter and extinction coefficients are commonly determined by the inversion of lidar signals by using the slope method, i.e., from a linear least-squares fit to the logarithm on the range-compensated lidar return. We investigate the accuracy of this method. A quantitative analysis is presented of the influence of white noise and atmospheric extinction on the accuracy of the slope method and on the maximum range of lidar systems. To meet this objective, we simulate lidar signals with extinction coefficients ranging from 10(-3) km(-1) to 10 km(-1) with different signal-to-noise ratios. It is shown that the backscatter coefficient can be determined by using the slope method with an ccuracy of better than ~ 10% if the extinction coefficient is smaller than 1 km(-1) and the signal-to-noise ratio is better than ~ 1000. The accuracy in the calculated extinction coefficient is only better than ~ 10% if the extinction is larger than 1 km(-1) and the signal-to-noise ratio is better than ~2000. If th atmospheric extinction coefficient is smaller than 0.1 km(-1), then it is not possible to invert the extinction from lidar measurements with an accuracy of 10% or better unless the signal-to-noise ratio isunrealistically high.