This study investigates the retrieval performances of LAI using radiative transfer model inversion. The SAIL and GEOSAIL models were used here to evaluate the impact of the complexity of canopy architecture on LAI estimation. The inversion of SAIL or GEOSAIL models was carried out using a look up table technique. Test data sets were generated with SAIL and GEOSAIL to evaluate the performances. Results indicate first that the way the solution is selected within the LUT appeared to be very important. Comparing the inversion using SAIL or GEOSAIL over test cases simulated with SAIL or GEOSAIL show that consistency between the RTM used in the inversion process and the test cases improves the estimation. However, estimation of LAI for complex canopy architecture with non random distribution of leaves (clumping) is difficult when no prior information is available on the targets. However, it is demonstrated that the effective LAI, i.e. the LAI retrieved from the gap fraction measurements assuming. random leaf distribution, is estimated with much better accuracy.