We present a time-domain, model-based scheme for X-ray induced acoustic tomography (XACT). In comparison to the commonly used analytic schemes, the model-based reconstructions exhibit less limited-view and noisy artifacts and enable incorporating detector sizes and sound speed variations. In this paper, we brie y compare the performance of the proposed model-based scheme with the universal back-projection and model back-projection reconstructions obtained for full- and limited view test-cases in spherical detection geometry for noisy acoustic measurements. The algorithm has also been validated for experimental ring-array XACT data.