The stochastic Liouville formalism (SLE) allows for electronic spin resonance (ESR) spectra to be computed for proteins and lipids, using a stochastic model for the motion of the nitroxide spin probe. The present work reports a molecular dynamics (MD) study of spin-labeled T4 lysozyme, where information extracted from the atomically detailed trajectories is used within the framework of the Liouville equation. Sets of 10 MD trajectories, with two force fields, were produced for each of the T4 mutants N40C and K48C, previously used in ESR experiments. Fluctuations of the local relevant degrees of freedom, order parameters for the spin probe, free induction decay profiles, and finally high-field ESR line shapes were calculated for the two proteins. Sampling probabilities for the nitroxide rotational degrees of freedom are in agreement with the traditional model of Brownian rotational diffusion in a restricting potential. Present results suggest that improved sampling (longer time scale, more averaging over initial conditions) is needed for a reliable multicomponent fitting to experimental spectra from molecular dynamics trajectories.