A T-fO2 pair is the result of solving two independent equations (for biotite and its host rock) each of which relates the two unknowns, oxygen fugacity (fO2) and temperature, with an analytically determined Fe2+/Fe3+ ratio. T-fO2 pairs, determined for a series of rock samples from localities which are spatially distributed over an intrusive are different and plot linearly in 1/T-logfO2 space. A comparison of such lines, estimated for granites from Australia, Japan and Portugal shows three major features: (1) magnetite-granites and magnetite-free ilmenite-granites cover distinct fields, (2) Australian granites have a lower oxidation state and higher biotite crystallization temperatures than granites from Japan, and (3) ilmenite-granites from Japan plot systematically in an oxygen-rich regime, that is above the NiNiO (NNO) buffer, implying that their mineral assemblage is not in equilibrium whereas ilmenite-granites from Australia plot in an oxygen-poor regime, i.e. below the fayalite-magnetite-quartz (FMQ) buffer. It is, of course, expected that magnetite-free ilmenite-granites with all Fe3+ partitioned into biotite, have an oxidation state below (FMQ). Ilmenite-granites which plot above (FMQ) must therefore have experienced oxidation upon cooling, after ilmenite and after biotite crystallization. Such an oxidation may be related to volatile water, available at temperatures below water saturation and above solidification of the rock. Because ilmenite-granites plot either below or above (FMQ) for a specific region, one may infer that this is related to a regionally low or high water content respectively. As such one can distinguish low-water regions, such as Australia and N Portugal which are recognized as non-subduction-related granites, and high-water regions, such as Japan and the Alps (Bergell) which are recognized as subduction-related granites. It is concluded that in non-subduction-related regimes the redox state of the source rock has been preserved in contrast to rocks in subduction-related regimes.