Enzyme analysis for Pompe disease in leukocytes has been greatly improved by the introduction of acarbose, a powerful inhibitor of interfering α-glucosidases, which are present in granulocytes but not in lymphocytes. Here we show that the application of acarbose in the enzymatic assay employing the artificial substrate 4-methylumbelliferyl-α-D-glucoside (MU-αGlc) is insufficient to clearly distinguish patients from healthy individuals in all cases. Also, the ratios of the activities without/with acarbose only marginally discriminated Pompe patients and healthy individuals. By contrast, when the natural substrate glycogen is used, the activity in leukocytes from patients (n = 82) with Pompe disease is at most 17% of the lowest control value. The use of artificial substrate in an assay with isolated lymphocytes instead of total leukocytes is a poor alternative as blood samples older than one day invariably yield lymphocyte preparations that are contaminated with granulocytes. To diagnose Pompe disease in leukocytes we recommend the use of glycogen as substrate in the presence of acarbose. This assay unequivocally excludes Pompe disease. To also exclude pseudo-deficiency of acid α-glucosidase caused by the sequence change c.271G>A (p.D91N or GAA2; homozygosity in approximately 1:1000 caucasians), a second assay employing MU-αGlc substrate plus acarbose or DNA analysis is required.