Cytospin preparations of endomyocardial biopsy washings were examined on 117 occasions from 13 heart transplant recipients and categorized according to the pattern of cell types observed. Twenty-nine percent of samples were acellular, a further 10% too bloodstained for analysis, and 61% were cellular. Eight lymphocytic samples were found and in all cases there was at least grade 1B rejection (four grade 1B, three grade 2, and one grade 3A) on histologic study. However, histologic study showed at least 1B rejection in 48% of cases when cytospins showed mixed inflammatory cells, 33% of cases when cytospins were histiocytic and in 35% when cytospins were bloodstained or acellular. Furthermore 16 of these rejection episodes with nonlymphocytic cytospins were grade 2. Although the recovery of a lymphocytic cytospin was specific for rejection, the sensitivity of the test was poor. Even when the sample is adequate, this method of biopsy washings will predict only one third of cases of significant acute rejection (grade 2 or worse). The large proportion of unsuitable samples also severely limits the utility of endomyocardial biopsy washings for the diagnosis of rejection. Histiocytic cytospins were seen in 63% of samples when previous biopsy sites were reported on histologic study and also in all three samples when histologic study showed ischemic injury. A mixed inflammatory cell pattern was seen to a lesser extent (31% of samples) in relation to previous biopsy sites. High peripheral blood lymphocyte counts were found when endomyocardial biopsy washings were lymphocytic or mixed inflammatory and also when histologic study showed endocardial lymphocytic infiltration (Quilty effect).