Abstract: In Lambda-grids, advance reservation is one of important methods to guarantee QoS for grid computing applications. However, it would bring many resource fragmentations, which have greatly effect on resource utilization. This paper discusses the key problems existing in advance reservation-based RWA algorithm, i.e., not considering resource fragmentation and its impacts to new arriving traffic request. Based on layered-graph model, a heuristic routing and wavelength assignment algorithm is proposed to support advance reservation with less resource fragmentations, named Less Resource Fragmentations based on Layered Graph (LG-LRF). The algorithm directly considers resource fragmentations as link weights of layered wavelength-plane graph, and therefore, can resolve routing and wavelength assignment at the same time, and avoid the wavelength continuity constraint and some side-effects on new arriving traffic request. Simulation results show that LG-LRF outperforms the conventional heuristic RWA algorithms in terms of blocking probability of traffic requests and resource fragmentations ratio, and is more suitable for advance reservation for lambda-grids.