We simultaneously determined global myocardial blood flow (MBF) by the argon inert gas technique and by nitrogen-13 ammonia positron emission tomography (PET) to validate PET-derived MBF values in humans. A total of 19 patients were investigated at rest (n = 19) and during adenosine-induced hyperaemia (n = 16). Regional coronary artery stenoses were ruled out by angiography. The argon inert gas method uses the difference of arterial and coronary sinus argon concentrations during inhalation of a mixture of 75% argon and 25% oxygen to estimate global MBF. It can be considered as valid as the microspheres technique, which, however, cannot be applied in humans. Dynamic PET was performed after injection of 0.8 +/- 0.2 GBq 13N-ammonia and MBF was calculated applying a two-tissue compartment model. MBF values derived from the argon method at rest and during the hyperaemic state were 1.03 +/- 0.24 ml min-1 g-1 and 2.64 +/- 1.02 ml min-1 g-1, respectively. MBF values derived from ammonia PET at rest and during hyperaemia were 0.95 +/- 0.23 ml min-1 g-1 and 2.44 +/- 0.81 ml min-1 g-1, respectively. The correlation between the two methods was close (y = 0.92x + 0.14, r = 0.96; P < 0.0001). No indication was found for limited extraction of ammonia in the myocardium. The high concordance of global MBF values derived with argon and ammonia indicates that the implicit correction of spillover and recovery effects, incorporated in the model by including an effective blood volume parameter, works correctly quantitatively. Our data provide the previously missing human validation of MBF measurements from 13N-ammonia PET.