Rock magnetic properties and palaeomagnetism of weakly metamorphosed banded-iron formations (BIFs) of the Palaeoproterozoic Hamersley Group, Western Australia, and Proterozoic BIF-derived iron ores have been investigated. The BIF units sampled here are slightly younger than 2500 Ma. At Paraburdoo, Mount Tom Price and Mount Newman iron ore formation was completed before 1850 Ma. Sampling was mainly from the Mount Tom Price and Paraburdoo mining areas and for the first time a palaeomagnetic fold test on fresh (unweathered and unaltered) BIF samples has allowed the nature of the remanence of the BIFs to be defined. The remanence of the BIFs is carried by late diagenetic/low-grade metamorphic magnetite after primary haematite. This remanence is pre-folding and is unlikely to be greatly affected by the high anisotropy because the palaeofield inclination was demonstrably low. Determination of palaeofield directions from measured remanence directions is complicated by self-demagnetization effects in strongly magnetic, highly anisotropic BIF specimens. We present a method for correcting measured directions for the effects of self-demagnetization and anisotropy. For typical BIFs, the effect of magnetic anisotropy on measured remanence inclinations and inferred palaeolatitudes is minor for low palaeolatitudes, but can lead to large errors in calculated palaeopoles for intermediate to moderately steep palaeolatitudes. Anisotropy also causes cones of confidence to be underestimated, due to compression of the range of inclinations. In principle, deflection of post-folding remanence towards the bedding plane by high magnetic anisotropy can produce an apparent syn-folding signature, with best agreement between directions from different fold limbs after partial unfolding. Thus high anisotropy cannot only bias estimated palaeofield directions and cause underestimation of errors, but can also mislead interpretation of the time of remanence acquisition. The anisotropy of anhysteretic remanent magnetization (ARM) probably yields an upper limit to the anisotropy of the chemical remanent magnetization (CRM) carried by the BIFs. Therefore, from the anisotropy of ARM, a maximum inclination deflection of 9° is suggested for the sampled BIFs. This corresponds to less than 5° change of palaeolatitude. The palaeomagnetic pole position calculated for BIFs at Paraburdoo is 40.9°S, 225.0°E (dp=2.9°, dm=5.8°) after tilt correction, but without correction for anisotropy. Other pole positions reported include that from flat-lying BIFs from Wittenoom at 36.4°S, 218.9°E (dp=4.6°, dm=9.1°), from Mount Tom Price iron ore at 37.4°S, 220.3°E (dp=5.7°, dm=11.3°) and from Paraburdoo ore at 36.4°S, 209.9°E (dp=4.7°, dm=8.8°). The poles from the BIFs, the Paraburdoo ore and the part of the Tom Price deposit that was sampled in this study are indistinguishable from each other and from the Mount Jope Volcanics overprint pole. The magnetization of the BIFs was probably acquired during burial metamorphism of the Hamersley Group, soon before the main folding and uplift event in the southern part of the Hamersley Province. This tectonic event exposed magnetite-rich BIFs to near-surface oxidizing conditions, producing extensive martite-goethite orebodies and also appears to have produced the syn-folding overprint magnetization recorded by the Mount Jope Volcanics of the underlying Fortescue Group. Ages of magnetization are tentatively interpreted as ∼ 2200±100 Ma for the BIFs, ∼ 2000±100 Ma for the supergene enrichment of BIF to martite-goethite ore, recorded by the Parabudoo and Mount Tom Price orebodies, and ∼ 1950±100 Ma for the metamorphic martite-microplaty haematite ore, recorded as an overprint by the Tom Price orebody and as the only surviving magnetization of the Mount Newman orebody.