The important functions of cell adhesion molecule L1 in the nervous system depend on diverse proteolytic enzymes which generate different L1 fragments. It has been reported that cleavage in the third fibronectin type III (FNIII) homologous domain generates the fragments L1-80 and L1-140, while cleavage in the first FNIII domain yields the fragments L1-70 and L1-135. These results raised questions concerning the L1 cleavage sites. We thus generated gene-edited mice expressing L1 with mutations of the cleavage sites either in the first or third FNIII domain. By immunoprecipitations and immunoblot analyses using brain homogenates and different L1 antibodies, we show that L1-70 and L1-135 are generated in wild-type mice, but not or only to a low extent in L1 mutant mice. L1-80 and L1-140 were not detected in wild-type or mutant mice. Mass spectrometry confirmed the results from immunoprecipitations and immunoblot analyses. Based on these observations, we propose that L1-70 and L1-135 are the predominant fragments in the mouse nervous system and that the third FNIII domain is decisive for generating these fragments. Treatment of cultured cerebellar neurons with trypsin or plasmin, which were both proposed to generate L1-80 and L1-140 by cleaving in the third FNIII domain, showed by immunoprecipitations and immunoblot analyses that both proteases lead to the generation of L1-70 and L1-135, but not L1-80 and L1-140. We discuss previous observations on the basis of our new results and propose a novel view on the molecular features that render previous and present observations compatible.