This thesis presents a non-linear model for the thermal and elastic properties of single-walled carbon nanotube reinforced polymer composites. Finite Element Analysis (1-EA), in conjunction with the Embedded Fiber Method (EFM), is used to calculate the effective stress-strain curve and thermal conductivity of the composite material. First, the geometry of a user-defined volume fraction of nanotubes is randomly generated and their properties are incorporated into the polymer matrix using the EFM. Non-linear FEA is next performed to account for the non-linear properties of the polymer matrix and the carbon nanotubes. Finally, Monte Carlo Analysis of five hundred random microstructures is performed to capture the stochastic nature of the random fiber generation and to derive statistically sound results. The model is validated by comparison with several different experiments reported in the open literature.