As a reliable alternative to autografts, decellularized peripheral nerve allografts (DPNAs) should mimic the complex microstructure of native nerves and be immunogenically compatible. Nevertheless, there is a current lack of decellularization methods able to remove peripheral nerve cells without significantly altering the nerve extracellular matrix (ECM). The aims of this study are firstly to characterize ex vivo, in a histological, biochemical, biomechanical and ultrastructural way, three novel chemical-enzymatic decellularization protocols (P1, P2 and P3) in rat sciatic nerves and compared with the Sondell classic decellularization method and then, to select the most promising DPNAs to be tested in vivo. All the DPNAs generated present an efficient removal of the cellular material and myelin, while preserving the laminin and collagen network of the ECM (except P3) and were free from any significant alterations in the biomechanical parameters and biocompatibility properties. Then, P1 and P2 were selected to evaluate their regenerative effectivity and were compared with Sondell and autograft techniques in an in vivo model of sciatic defect with a 10-mm gap, after 15 weeks of follow-up. All study groups showed a partial motor and sensory recovery that were in correlation with the histological, histomorphometrical and ultrastructural analyses of nerve regeneration, being P2 the protocol showing the most similar results to the autograft control group.