In the past few years, clinical trials with multielectrode "basket" catheters have provided unique recordings from the endocardium of intact in-situ human hearts. Analysis of these recordings is difficult because unipolar electrograms obtained from a basket catheter in the blood-filled cavity differ from those obtained by other mapping techniques such as endocardial balloons used during antiarrhythmic surgery. We investigated these differences using basket catheter recordings obtained in isolated porcine and canine hearts that could be filled with perfusion fluid and evacuated at will. The results indicated that the differences between basket and balloon recordings are largely attributable to the presence and absence of blood. Activation maps obtained in the presence and absence of blood were usually similar and only differed in a minority of cases at sites in which electrograms revealed multiple deflections. In conclusion, unipolar mapping using a basket catheter can be used with confidence for the creation of activation maps if appropriate care is taken in the interpretation of fractionated electrograms.