Abstract: A method to extract the basic information from laser altimeter return waveforms is developed, by which decompose a laser altimeter return waveform into a series of Gaussian components and obtain the number of the Gaussian components and the position, half-width and amplitude of each Gaussian component. From these basic parameters, the topographic information as the elevations and distributions, the surface slope and roughness, and the albedos of distinct reflecting surfaces within the laser footprint can be further induced. Firstly, the initial values of the number of Gaussian components and the position, half-wide of each component are estimated from the number and positions of the waveform's inflection points. Secondly the linear least-squares method are used to calculate the initial amplitudes of each Gaussian component. At last the waveform is fitted with the chosen and ranked Gaussians by Levenberg-Marquardt method, achieving basic optimized parameters of Gaussians. Three types of simulated return waveforms of the laser altimeter from scalariform terrain, woodland and slope surface have been fitted by this decomposing algorithm, the original waveforms are successfully approximated by the fitted ones and the optimized parameters of simulated return waveforms are satisfied. And the results prove that the decomposing algorithm of laser altimeter waveforms obtains a higher precision, it will have a good effect on the analysis of laser altimeter return waveforms.