Early-diagenetic REE-phosphate minerals are widespread, although minor (<0.005-0.14 wt percent), constituents of Archaean to Cretaceous sandstones from Australian sedimentary basins. Authigenic florencite ((REE)Al-3(PO4)(2)(OH)(6)), crandallite (CaAl3(PO4)2(OH)(6)), and gorceixite (BaAl3(PO4)(2)(OH)(6)) occur as clusters of minute crystals (<0.1-10 mu m in width) within pockets of clay matrix lining detrital quartz surfaces and less commonly within altered mica and feldspar grains. Rare florencite crystals form along the surface of detrital monazite grains, and in one locality, crandallite and gorceixite crystals line cavities left after grain dissolution. The aluminophosphate minerals commonly display compositional zoning, with major cation interchange between REE, Sr2+, Ca2+, and Ba2+, and minor anion interchange between PO43-, SO42-, and CO32-. Xenotime (YPO4), which contains minor amounts of Th, U, and REE, forms pyramidal, dentate overgrowths (<1-20 mu m) on detrital zircon grains and locally partly surrounds detrital quartz surfaces. Authigenic apatite is generally less abundant than diagenetic aluminophosphate and occurs as minute, pore-filling and grain-lining crystals. The authigenic phosphate minerals precipitated shortly after burial within the zone of sulfate reduction and methanogenesis. The phosphates probably formed from P and REE released into marine sediment pore waters following bacterial decomposition of organic matter and reduction of hydrous iron oxides, as well as from partially dissolved and altered detrital minerals (monazite, clay matrix, feldspar, and mica). For the aluminophosphates, the critical factor controlling their precipitation site is the availability of Al (and to a lesser extent adsorbed REE and P), whereas xenotime crystals mostly Precipitate on isomorphous substrates (for example, detrital zircon grains). Given the widespread occurrence of early-diagenetic aluminophosphates and xenotime in Australian marine-deposited sandstones, it seems surprising that they have not been noticed elsewhere. However, their minute crystal size (<0.1-10 mu m) and low concentration (generally <0.05 wt percent) probably hindered their previous identification by conventional microscopic techniques, while their marked insolubility at low temperatures rendered them inert to sequential solvent extraction geochemistry. Accordingly, these minerals represent a previously unrecognized marine sink for reactive P. Estimates from this study show that the global flux of oceanic P removed through the precipitation of the REE-phosphates (conservatively 7.56 x 10(10) moles yr(-1)) may be of equal importance to other major sinks (for example, carbonate fluorapatite, organic-P). If so, then the total reactive P output is probably considerably greater than previous estimates and, for the oceanic P cycle to remain In a steady state, implies a decrease in the output of other sinks or an increase In the P input, A higher output of reactive P will also reduce estimates of the oceanic residence time of P (from this study, between 11-20 ka), As P is crucial for sustaining marine life, such changes in the P budget of the oceans may affect estimates of the biomass of marine organisms the biotic fixation of CO2, and therefore predictions of global climatic change.