Although many lysosomal disorders are corrected by a small amount of the missing enzyme, it has been generally accepted that 20–30% of normal acid α-glucosidase (GAA) activity, provided by gene or enzyme replacement therapy, would be required to reverse the myopathy and cardiomyopathy in Pompe disease. We have addressed the issue of reversibility of the disease in the Gaa^–/– mouse model. We have made transgenic lines expressing human GAA in skeletal and cardiac muscle of Gaa^–/– mice, and we turned the transgene on at different stages of disease progression by using a tetracycline-controllable system. We have demonstrated that levels of 20–30% of normal activity are indeed sufficient to clear glycogen in the heart of young Gaa^–/– mice, but not in older mice with a considerably higher glycogen load. However, in skeletal muscle—a major organ affected in infantile and in milder, late-onset variants in humans—induction of GAA expression in young Gaa^–/– mice to levels greatly exceeding wildtype values did not result in full phenotypic correction, and some muscle fibers showed little or no glycogen clearance. The results demonstrate that complete reversal of pathology in skeletal muscle or long-affected heart muscle will require much more enzyme than previously expected or a different approach.