—Cardiac myocytes are electrically coupled by gap junctions, clusters of low-resistance intercellular channels composed of connexins. Variations in the quantities and spatial distribution of different connexin types have been implicated in regional differentiation of electrophysiological properties in the heart. Although independent studies have demonstrated that connexin43 is abundant in working ventricular myocardium and that connexin40 is preferentially expressed in the atrioventricular conduction system of a number of species, information on the spatial distribution of connexin45 in the heart is limited to data obtained using an antibody raised to a single peptide sequence. In the present study, we report on the production and characterization of a new anti-connexin45 antibody and its application to the investigation of connexin45 expression in mouse and rat myocardium. The affinity-purified antiserum, raised in guinea pig to residues 354 to 367 of human connexin45, recognized a single 45-kD band on Western blots of HeLa cells transfected to express connexin45 and gave punctate immunolabeling at the cell borders, demonstrated by freeze-fracture cytochemistry to represent gap junctions. Only low levels of connexin45 mRNA were detected on Northern blots of mouse and rat cardiac tissues, and connexin45 protein levels were below the limit of detection on Western blots. Confocal microscopy of immunolabeled ventricular tissue revealed that the major part of the working myocardium was immunonegative for connexin45. A clearly defined zone containing connexin45-expressing cells was, however, localized to the endocardial surface, overlapping with connexin40-expressing myocytes of the conduction system. As these results contrast with the prevailing view that connexin45 is widely distributed in working ventricular myocytes, we compared the immunolabeling pattern obtained with a commercially supplied anti-connexin45 antiserum raised against the same peptide that was used in previous studies. The commercial connexin45 antiserum gave widespread labeling throughout the ventricular myocardium, but this labeling was inhibited by a six–amino acid peptide matching part of the connexin43 sequence, indicating cross-reaction of the commercial connexin45 antiserum with connexin43 in the tissue. Further evidence for such cross-reactivity came from observations on connexin43-transfected cells, which gave positive immunolabeling with the commercial anti-connexin45 antiserum. Our demonstration of a specific association of connexin45 with connexin40-expressing myocytes in rat and mouse ventricle raises the possibility that connexin45 contributes to the modulation of electrophysiological properties in the ventricular conduction system and highlights the need for reappraisal of the distribution and role of connexin45 in other species.