For an adequate evaluation of mitochondria from diseased hearts, basic characteristics of isolation, storage, media, ultrastructure and type of assay were first determined using mitochondria from normal animals. A proteinase procedure yielded mitochondria from small laboratory animals, with low respiratory control and marked permeability changes. The isolation medium yielding the most stable mitochondria with the highest respiratory control contained 0.18M KCl, 10mM EDTA, and 0.5% to 1% bovine serum albumin at pH 7.2. Heart failure in guinea pigs and rabbits was produced by varying degrees of stenosis of the ascending aorta. An aberration in respiratory control was found in mitochondria from hearts in severe failure. The quantitative differences between normal and experimental respiratory control values were greatest when the highest possible normal respiratory control levels were obtained. The difference between mitochondria prepared by a proteinase method from control and failing hearts was minimal. No changes in oxidative phosphorylation were noted in mitochondria from hearts arrested by nitrogen, suggesting that acute hypoxia does not irreversibly damage energy-liberating reactions. It is concluded that severe heart failure is characterized by defects in mitochondrial oxidative phosphorylation, and that techniques of isolation or assay or both are probably not causing the abnormalities.