Aims: The fast sodium channel (FSC) plays an important role in the formation of cardiac action potentials (APs). Dysfunction of FSC underlies many severe pathologies, since it activates first in an AP and leads the whole process of cell activation and various downstream cell behaviours. However, models of FSC adopted in recent mouse cell models are relatively outdated: there has been no major modification on Hudgkin and Huxley (HH) formulated FSC models since 1990s, which was fitted to data from other species; newer Markov chain models were developed more than 15 years ago and only for ventricles. Therefore, in this modelling study, we aim at fitting new FSC models with recently published experimental data on mouse atrial and ventricular FSCs. Methods: Based on the existing and widely used two types of FSC models (with HH formulation and Markov chain), we refitted models of mouse atrial and ventricular FSCs with new experimental data. Automated and manual fitting techniques were combined to get satisfied results. Markov chain models were constructed as simple as possible to reduce the computational requirements. Then the models were compared with old ones and incorporated into a mouse ventricular cell model and a newly developed mouse atrial cell model in order to measure their influences on APs. Results: First, taking into consideration of the electrophysiological heterogeneity between mouse atria and ventricles, we acquired several new mouse FSC models which better reproduced new experimental findings. Second, we found in our simulation that the FSC in mouse atria with higher maximum conductance and left shifted activation gate significantly affected the excitability and activation process of the mouse atrial cell, making it easier to be activated and causing instability of APs.