In 1964, Christian de Duve first suggested that enzyme replacement might prove therapeutic for lysosomal storage diseases (LSDs). Early efforts identified the major obstacles, including the inability to produce large quantities of the normal enzymes, the lack of animal models for proof-of-concept studies, and the potentially harmful immune responses to the "foreign" normal enzymes. Subsequently, the identification of receptor-mediated targeting of lysosomal enzymes, the cloning and overexpression of human lysosomal genes, and the generation of murine models markedly facilitated the development of enzyme replacement therapy (ERT). However, ERT did not become a reality until the early 1990s, when its safety and effectiveness were demonstrated for the treatment of type 1 Gaucher disease. Today, ERT is approved for six LSDs, and clinical trials with recombinant human enzymes are ongoing in several others. Here, we review the lessons learned from 20 years of experience, with an emphasis on the general principles for effective ERT and the remaining challenges.