The sawing of tropical hardwood species is a big challenge for sawmills in developing countries. In order to investigate sawing strategies and volume recovery of tropical hardwood species, a log shape database was created using a portable 3D laser scanner. The data were collected in Mozambique, where twelve Jambirre (Millettia stuhlmannii Taub.) and five Umbila (Pterocarpus angolensis DC) logs were scanned. The logs were selected among the most commercialized species and the crook was the main selection parameter. In addition, straight logs were incorporated as reference. A saw simulation Matlab algorithm that combines skew and rotation was developed. The results show that point cloud data from the 3D scanner provide detailed models of the external log geometry and accurately describe the log shapes and volumes. Preliminary results from breakdown simulation revealed that the through-and-through sawing pattern yields more than the cant saw pattern and that the increase in yield was almost the same for both species.