Estimating forest canopy height from large-footprint satellite LiDAR waveforms is very challenging given the complex interaction between the LiDAR waveforms, the terrain, and the vegetation especially in dense tropical and equatorial forests. From the launch of GLAS (Geoscience Laser Altimeter System) onboard ICESat (Ice, Cloud, and land Elevation Satellite), numerous studies developed statistical models to estimate canopy heights from GLAS waveforms using metrics extracted from these waveforms as well as additional data extracted from Digital Elevation Models (DEM). In this study, several multiple linear regression models were first performed between the canopy height and different GLAS and SRTM metrics over French Guiana. Next, Random Forest technique was used using the same GLAS and SRTM (Shuttle Radar Topography Mission) DEM metrics. Finally, a regression model after principal component analysis of GLAS waveforms was evaluated and compared to other tested models.