Plant fiber properties, such as morphological and mechanical, are characterized by a large dispersion. Thereby, a statistical analysis is needed to obtain consistent results. An experimental study, conducted on 50 flax yarns, shows that the flax yarn properties (Young's modulus, tensile strength and diameter) follow Gaussian distributions. This approach is obviously reliable, however time-consuming, to get relevant information. An alternative could be the identification of the yarn mechanical properties using an inverse approach, based on tensile tests conducted on flax fabric reinforcement. The aim of this study is to develop a numerical method that allows to identify the statistical distributions of flax yarn properties based on tensile tests conducted on fabric specimens. The proposed strategy relies on two assumptions. On the one hand, fabric is constituted of several yarns acting like springs in parallel. On the other hand, yarns are considered as brittle-elastic materials. Hence, a yarn breaks when the load reaches its failure strength, leading to a loading redistribution to the intact yarns. A comparison of the numerical and experimental results of flax fabric tensile behavior, and of the flax property statistical distributions allows to confirm the performance of this strategy. The results show that the flax fabric tensile behavior is correctly described by the proposed modeling strategy. The average and the standard deviation of the Young's modulus, the tensile strength and the diameter of flax yarns identified from fabric tensile tests via inverse approach are close to those obtained experimentally on individual yarns. Indeed, the measured average diameter of unitary yarns are 244.4 ± 19.2 μm and the fitted one is 2474.8 ± 17.4 μm, the fitting failure strength 292.3 ± 33.4 MPa are close to the experimental 271.2 ± 47.5 MPa. The identified Young's modulus is 9.4 ± 0.9 GPa is lower than the experimental 10.8 ± 1.3 GPa.