In 1956, Verhoogen proposed that ionic reordering during low-temperature oxidation of titanomagnetite could result in self-reversal of natural remanent magnetization. Later, in 1966, O'Reilly and Banerjee concluded that (1) the compositional field of self-reversing titanomaghemite was restricted to very high oxidation states and (2) that high oxidation temperature (≥250°C) would be required to reach these oxidation levels, which would in turn be inconsistent with stability of the cation-deficient structure of titanomaghemite in nature. However, an example of the process has been reported, in some altered pillow basalt lavas of Late Cretaceous age from Detroit Seamount (Ocean Drilling Program (ODP) sites 883, 1203, and 1204) in the northwestern Pacific Ocean. Here we present new paleomagnetic, rock magnetic and compositional data from mid-ocean ridge tholeiitic basalts of Late Jurassic age collected at Deep Sea Drilling Project (DSDP) Site 307. Although severely oxidized, the DSDP Site 307 basalt samples lack the clear self-reversed magnetizations seen in some samples from Detroit Seamount. These data constrain the compositional field of self-reversing titanomaghemite to extremely high oxidation states (z ≥ 0.95), consistent with the first conclusion of O'Reilly and Banerjee. The comparison of results from ODP sites 883, 1203, and 1204 on Detroit Seamount and those from the DSDP Site 307 basalts suggests that the presence of self-reversed magnetizations may signify unusually vigorous and sustained fluid flow that resulted in very efficient removal of Fe from titanomaghemite producing the extremely high oxidation states required for self-reversal.