Mycophenolate mofetil is used increasingly to provide immunosuppression after nonmyeloablative allogeneic hematopoietic cell transplantation. There is wide variability in the pharmacokinetics of mycophenolic acid (MPA), the active metabolite, and low concentrations are associated with rejection after organ transplantation. We hypothesized that low MPA was associated with poorer engraftment and a higher incidence of acute graft versus host disease. We evaluated the pharmacokinetics in 87 adult subjects undergoing nonmyeloablative-related and nonmyeloablative-unrelated hematopoietic cell transplantation who were receiving 1 g mycophenolate mofetil orally or intravenously every 12 hours plus cyclosporine (INN, ciclosporin). Subjects with an unbound MPA area under the curve (AUC) from 0 to 6 hours of less than 150 ng·h/mL had a higher cumulative incidence of grade II-IV acute graft versus host disease than subjects with a greater AUC (68% versus 40%, P = .02). An unbound AUC from 0 to 12 hours of less than 300 ng·h/mL was also associated with more frequent acute graft versus host disease (58% versus 35%, P = .05). There was no association between graft versus host disease and trough concentrations (P ≤ .62). A higher cumulative incidence of engraftment was associated with total MPA trough concentrations greater than 1 μg/mL (P < .01). All engraftment failures occurred in the cord blood recipients. About one half of subjects were below the unbound AUC target after oral dosing with nearly a 5-fold variability in AUC. Intravenous dosing achieved unbound targets better than oral dosing. The current practice of dosing with 1 g twice daily provides inadequate plasma concentrations in many patients, and doses of at least 3 g/d are likely necessary. Therapeutic monitoring of MPA concentrations with dose adjustment into the therapeutic target appears to be necessary for the most effective use of mycophenolate mofetil.