One of the most promising applications of hydrolytically degrading biomaterials is their use as drug release carriers. These uses, however, require that the degradation and diffusion of drug are reliably predicted, which is complex to achieve through present experimental methods. Atomistic modeling can help in the knowledge-based design of degrading biomaterials with tuned drug delivery properties, giving insights on the small molecules diffusivity at intermediate states of the degradation process. We present here an atomistic-based approach to investigate the diffusion of water (through which hydrolytic degradation occurs) in degrading bulk models of poly(lactic acid) or PLA. We determine the water diffusion coefficient for different swelling states of the polymeric matrix (from almost dry to pure water) and for different degrees of degradation. We show that water diffusivity is highly influenced by the swelling degree, while little or not influenced by the degradation state. This approach, giving water diffusivity for different states of the matrix, can be combined with diffusion-reaction analytical methods in order to predict the degradation path on longer time scales. Furthermore, atomistic approach can be used to investigate diffusion of other relevant small molecules, eventually leading to the a priori knowledge of degradable biomaterials transport properties, helping the design of the drug delivery systems.