Abstract Background: Mitochondrial neurogastrointestinal encephalomyopathy (MNGIE) is caused by mutations in the gene encoding thymidine phosphorylase (TP). The clinical manifestations of MNGIE are recognizable and homogeneous, but in the early stages, the disease is often misdiagnosed. This study assesses the reliability of biochemical assays to diagnose MNGIE. Methods: We studied 180 patients with clinical features suggestive of MNGIE, 14 asymptomatic TP mutation carriers, and 20 controls. TP enzyme activity in the buffy coat was determined by a fixed-time method, and the plasma nucleosides thymidine (dThd) and deoxyuridine (dUrd) were assessed by a gradient-elution reversed phase HPLC method. TP was sequenced through standard procedures in patients who met the clinical criteria for MNGIE. Results:Twenty-five of the 180 patients fulfilled the clinical criteria for MNGIE and had homozygous or compound heterozygous TP mutations. All had drastically decreased TP activity [mean (SD), 10 (15) nmol thymine formed · h−1 · (mg protein)−1 vs 634 (217) nmol thymine formed · h−1 · (mg protein)−1 for the controls]. Relative to the control mean, TP activities were reduced to 35% in mutation carriers and 65% in MNGIE-like patients. All 25 MNGIE patients had detectable plasma dThd [8.6 (3.4) μmol/L] and dUrd [14.2 (4.4) μmol/L]. Controls, carriers, and MNGIE-like patients showed no detectable plasma dThd and dUrd. Conclusions:We propose a diagnostic algorithm based on the determination of plasma dThd and dUrd, TP activity in buffy coat, or both to make a definitive diagnosis of MNGIE. Increased concentrations of dThd (>3 μmol/L) and dUrd (>5 μmol/L) in plasma or a decrease in buffy coat TP activity to ≤8% relative to controls is sufficient to diagnose MNGIE.