The potential of obidoxime and other pyridinium-4-aldoximes to reactivate dimethyl- and diethylphosphorylated cholinesterases is markedly restricted by the inevitable formation of rather stable phosphoryl oximes (POXs) with high anticholinesterase activity. This effect is hardly seen with very dilute enzyme preparations, but becomes significant at physiological enzyme concentrations. Human plasma with the butyrylcholinesterase irreversibly blocked by soman was able to stimulate obidoxime-induced reactivation of concentrated erythrocyte acetylcholinesterase (Ery-AChE) to the same extent as was observed with a dilute preparation, suggesting phosphoryl oxime-destroying capacity. The inactivating factor, which was tentatively termed POX-hydrolase, had (1) a molecular weight of >100 kDa; (2) required Ca2+ , which could not be substituted by Zn2+ or Mg2+; and (3) lost its catalytic activity reversibly in the presence of ethylenediamine-tetraacetic acid (EDTA). The enzyme activity varied widely (20-fold) among different subjects and did not follow the activity pattern of human serum paraoxonase (PON1). Rabbit plasma with its particularly high paraoxonase content showed only weak POX-hydrolase activity. These data suggest POX-hydrolase to be a different entity. POX-hydrolase was most active with the putative phosphoryl-obidoxime from paraoxon-ethyl, less with the product from paraoxon-methyl and least with that from diisopropylfluorophosphate. The analogue TMB-4 reacted similarly to obidoxime. The putative phosphonyl oximes arising by the reaction of obidoxime with nerve agents were apparently not cleaved. The variation in POX-hydrolase activity may additionally contribute to the variable response to oxime therapy in patients with organophosphate insecticide poisoning.