Ninety-three patients with PTE (that is, hematocrit 51% or greater) were identified among 431 renal transplant recipients, an incidence of 21.6%. Thirty-eight patients underwent blood volume measurements, and 22 of these had high red cell volume and therefore were considered to have true PTE. To analyze factors predictive of erythrocytosis, a control group with normal hematocrit was randomly selected from our renal transplant population and compared with the 93 patients with PTE, and with the 22 who had true PTE. Using step-wise logistic regression analysis, we identified three variables that were consistent predictors of PTE. In order of significance, the serum creatinine value at the onset of PTE appears to most strongly predict the occurrence of PTE (P less than 0.0001). As creatinine value increases, the probability of PTE decreases. Next was immunosuppression, where double immunosuppressive therapy was associated with a greater probability of PTE than triple therapy (P less than 0.0001). The overall incidence of PTE in patients on double therapy was 34%, while that for those on triple therapy 10.4%. Last was duration of dialysis for which increasing values correspond to increasing probability of PTE (P = 0.004). Comparison of the serum erythropoietin (EPO) levels for patients and controls yielded a nonsignificant result (P = 0.2507 and P = 0.383 for all patients with PTE and true PTE, respectively), and therefore EPO levels were inappropriately elevated for the level of hematocrit in the PTE group. Only the number of rejections and duration of follow-up (r = -0.3507) were significantly correlated with EPO (P less than 0.05).(ABSTRACT TRUNCATED AT 250 WORDS)