The mechanical and damage properties of individual constituents used in the fibrous composites are very essential in the micromechanical damage modeling. As fibres are the principal load carrying component of a composite, the knowledge of statistical strength distribution of fibres is very essential for estimating the effective properties of the composite and for its damage investigation. Even though various micromechanical models are available in the literature, most of them disregard the effects of diametric irregularities and variation in properties of the fibres. This may be attributed to the discrepancy between theoretical predictions and experimental results. So, in order to model the uncertainties in composites, the statistical strength distribution of the single fibres should be known a priori. In this work various damage parameters like ultimate strength and failure strains of Torayca R T300 carbon fibres are experimentally characterized for statistical analysis and various probabilistic methods are studied to predict the life data of the fibre in terms of measured stresses. The statistical strength variability of single fibres is described using the Weibull distribution and the probability of failure is predicted. This probabilistic model is developed based on the weakest link hypothesis. Here, the focus of the work is on the reliability characterization of single fibres using a generalized modified four parameter Weibull model having stochastic variables which accounts for geometric irregularities and variation in properties. Finally, the probabilistic estimation of tensile strength given by these models are compared with the experimental results. © 2015, American Institute of Aeronautics and Astronautics. All Rights Reserved.