Background: PR-104 is a phosphate ester that is systemically converted to the corresponding alcohol PR-104A. The latter is activated by nitroreduction in tumours to cytotoxic DNA cross-linking metabolites. Here, we report a population pharmacokinetic (PK) model for PR-104 and PR-104A in non-human species and in humans. Methods: A compartmental model was used to fit plasma PR-104 and PR-104A concentration-time data after intravenous (i.v.) dosing of humans, Beagle dogs, Sprague-Dawley rats and CD-1 nude mice. Intraperitoneal (i.p.) PR-104 and i.v. PR-104A dosing of mice was also investigated. Protein binding was measured in plasma from each species. Unbound drug clearances and volumes were scaled allometrically. Results: A two-compartment model described the disposition of PR-104 and PR-104A in all four species. PR-104 was cleared rapidly by first-order (mice, rats, dogs) or mixed-order (humans) metabolism to PR-104A in the central compartment. The estimated unbound human clearance of PR104A was 211 L/h/70 kg, with a steady state unbound volume of 105 L/70 kg. The size equivalent unbound PR-104A clearance was 2.5 times faster in dogs, 0.78 times slower in rats and 0.63 times slower in mice, which may reflect reported species differences in PR-104A O-glucuronidation. Conclusions: The PK model demonstrates faster size equivalent clearance of PR-104A in dogs and humans than rodents. Dose-limiting myelotoxicity restricts the exposure of PR-104A in humans to approximately 25 of that achievable in mice.