We have developed a new automated method for the determination of malate dehydrogenase (MDH) isoenzymes in serum employing guanidine hydrochloride. Our proposed method showed good reproducibility; within-run precision coefficient of variations (CVs) were less than 2.5 (mean 13.6-42.9 U/L) for total MDH (T-MDH) and less than 6.7% (mean 6.3-23.5 U/L) for mitochondrial MDH (m-MDH) (n = 10). The upper detection limit of the proposed method exhibited good linearity up to 1,000 U/L for both T-MDH and m-MDH. In the proposed m-MDH reagent, the presence of up to 2,000 U/L of cytosolic MDH(c-MDH) activity had no effect on the outcome of m-MDH assay. Results of our proposed method (y) correlated well with those of the electrophoretic method (x) giving the regression equation: y = 1.46 x + 6.87 (N = 30); r = 0.99. Normal concentrations of various anticoagulants and bilirubin did not affect the assay results. Both ascorbic acid and glucose exhibited a slight positive interference with the proposed assay. Clinically, we found that m-MDH activity in serum had greater diagnostic predictive value than T-MDH activity for judging successful outcome of reperfusion therapy; the prognosis was poor when the m-MDH/T-MDH ratio was greater than 20%.