Staining with triphenyltetrazolium chloride (TTC), although controversial, has frequently been used for the delineation of myocardial infarction. This study was performed further to explore the reliability of the TTC method. In 24-h experiments pigs were subjected to closed-chest occlusion of the left anterior descending coronary artery for 30, 60 or 90 min followed by reperfusion with or without superoxide dismutase (SOD) as an adjunct. One TTC-stained slice from each heart was stabilized by microwave irradiation, gelatin-embedded, frozen in hexane chilled with dry ice and cryosectioned. Serial sections were stained with antibodies against fibronectin in order to identify irreversibly injured myocytes and with van Gieson histologically to confirm the necrotic tissue. A close correspondence of the infarct size was found between TTC stained slices and anti-fibronectin stained sections. The infarct size in the van Gieson stained sections also showed good correspondence but the area of infarction tended to be larger. In the experimental group subjected to 30 min ischaemia and with SOD as an adjunct, the estimated infarcted area in the TTC stained slices was significantly smaller than the area estimated from the anti-fibronectin stained sections. In sections viewed in the light microscope an inverse pattern of TTC and anti-fibronectin staining was observed. It was confirmed at the light microscopic level that myocytes containing an abundance of TTC deposits lacked fibronectin whereas myocytes stained with antifibronectin in general lacked TTC staining except for a zone approximately 0.5 mm wide which was located at the intersection between damaged and surviving myocytes where small TTC deposits were present. The width of the stained zone did not differ among the experimental groups. Thus, differences in estimated infarct size by the three methods used reflect problems in correctly delineating the border between living and dead myocardium rather than an interference by SOD on TTC staining.