cis-4-Hydroxycyclophosphamide (2) and aldophosphamide (4) were generated in aqueous phosphate or cacodylate buffer by dimethyl sulfide reduction of cis-4-hydroperoxycyclophosphamide (8) and by sodium periodate cleavage of 3,4-dihydroxybutyl N,N-bis(2-chloroethyl)phosphorodiamidate (9), respectively; the reactions of 2 and 4 were examined by 1H and 31P NMR. Within 30-60 min (pH or pD 7.0, 25 °C) the same pseudoequilibrium mixture was established in both reactions, with cis-and trons-4-hydroxycyclophosphamide (2 and 3), aldophosphamide (4), and its hydrate (5) present in the approximate ratio of 4:2:0.3:1. Structures of the intermediates were assigned unambiguously based upon analysis of the chemical shifts and coupling constants in the proton spectra determined in D2o buffers, and the 31P assignments followed by correlation of component ratios at equilibrium. Free energy differences of 0.4,0.4, and 0.7 kcal/mol at 25 °C were estimated between 2, 3, 5, and 4, respectively, with 2 being the most stable. The aldehyde 4 reacted most rapidly with water to give hydrate 5; cyclization of 4 to 3 occurred faster than to 2, and the rate of cyclization to 2 was comparable to that for elimination to 6. Compound 5 is formed much faster than 3 from the diol cleavage, but 5 and 3 are produced at comparable rates from 2, suggesting that conversion of 2 to 3 can proceed by a mechanism other than ring opening. The rate of equilibration appears to be independent of buffer structure, indicating that bifunctional catalysis is not important in the ring-opening reaction. β-Elimination from 4 is rate limiting for the production of 6 and acrolein, and the rate for phosphate is 2-to 3-fold faster than for cacodylate under identical conditions. These results provide the first definitive evidence for the stability of the elusive aldehyde 4 in aqueous solution and for the existence of a preequilibrium among 2-5 prior to rate-limiting expulsion of phosphoramide mustard from 4.