Current published data on therapeutic leukapheresis in hyperleucocytic AML does not define the impact on survival from this procedure. Between 1992 and 1999 we saw 146 patients with newly-diagnosed AML (APL excluded) and an initial WBC count > 50 x 10(9)/L of whom 71 underwent leukapheresis at the discretion of their treating doctors. We compared outcome (early mortality, CR, and overall survival) rates in the patients who were and were not pheresed. After accounting for covariates relevant to these outcomes, including age, performance status, and cytogenetics, there was evidence (p = .006) that pheresis reduced 2-week mortality rate and a suggestion (p = .06) that this resulted in a higher CR rate. However there was no evidence that pheresis lengthened longer-term or overall survival; if anything the suggestion was the converse (p = .06). These data may reflect the fact that the patients chosen to have pheresis were prognostically unfavorable as defined by variables that were not captured in our data set, since the alternative explanation i.e. that pheresis per se shortens overall survival seems less likely. Whether the above justifies the use of pheresis in the absence of evidence from a randomized trial is doubtful, but it seems likely that any long-term benefit to be derived from this procedure must await further advances in anti-leukemia therapy.