Use of radiolabeled microspheres is a standard method to measure regional myocardial perfusion in animals. Human albumin microspheres have been given safely to patients, but positron-emitting 68Ga-labeled human albumin microspheres are characterized by an unstable radiolabel. A new labeling procedure that covalently binds 11C (t1/2 = 20.3 min) to human albumin microspheres via 11CH3I was developed. Seven open-chest and two closed-chest dogs were studied. Reference and 11C-labeled human albumin microspheres (2 to 25 mCi) were both injected into the left atrium. Positron tomographic images were obtained of the myocardial distribution of the 11C-labeled microspheres. Timed arterial withdrawal was used for both reference gamma-labeled microspheres and 11C-labeled human albumin microspheres. Myocardial tissue samples matched to tomographic slices were well-counted for calculation of reference values of regional myocardial perfusion. Serial venous blood samples for residual 11C activity of 30 and 60 min after injection were less than 1% of the myocardial 11C concentration demonstrating a stable 11C bond to the human albumin microspheres. Regional myocardial perfusion calculated by this technique correlated well with values obtained with reference microspheres (r = .97) over a range of 0.2 to 3.5 ml/min/g. Correction for wall thickness improved the slope of the regression equation from y = 0.71 x -0.03 to y = 0.88 X -0.05. Thus, 11C human albumin microspheres are stable radiochemically and can be used as a quantitative measure of regional myocardial perfusion.