Carbon nanotubes can be randomly deposited in polymer thin film matrices to form nanocomposite strain sensors. However, a computational framework that enables the direct design of these nanocomposite thin films is still lacking. The objective of this study is to derive an experimentally validated and two-dimensional numerical model of carbon nanotube-based thin film strain sensors. This study consisted of two parts. First, multi-walled carbon nanotube (MWCNT)-Pluronic strain sensors were fabricated using vacuum filtration, and their physical, electrical, and electromechanical properties were evaluated. Second, scanning electron microscope images of the films were used for identifying topological features of the percolated MWCNT network, where the information obtained was then utilized for developing the numerical model. Validation of the numerical model was achieved by ensuring that the area ratios (of MWCNTs relative to the polymer matrix) were equivalent for both the experimental and modeled cases. Strain sensing behavior of the percolation-based model was simulated and then compared to experimental test results.