The Red Clay unit of northern China is a widespread Miocene–Pliocene (7.2–2.5 Ma) aeolian deposit that underlies much of the Quaternary loess–paleosol sequence in the central Loess Plateau. It comprises a sequence of highly developed soils and interbedded layers of less weathered loess-like material, here referred to as reddish loess. In common with the loess–paleosol alternations of the overlying Quaternary deposits, the soil–reddish loess alternations are interpreted to represent climatic fluctuations, in this case between warm-humid and relatively dry-cool conditions. However, while magnetic susceptibility variations in the Quaternary deposits provide a good proxy for pedogenic intensity, in the Red Clay the relationship is less clear and magnetic susceptibility data are in conflict with other measures of pedogenesis. In an attempt to resolve this issue, we have investigated a Red Clay section and overlying loess–paleosol sequence at Xifeng and have supplemented these data with samples from the Red Clay at Lingtai. Our study indicates that the Red Clay and loess–paleosol sequence have a common magnetic mineralogy comprising magnetite, maghemite and hematite (and possibly goethite), and that both sequences show a good correlation between magnetic susceptibility and the magnitude of the superparamagnetic (SP) component. This implies that susceptibility enhancement in the Red Clay is strongly related to the magnitude of the SP content, a component that is widely recognised as having a pedogenic origin in Quaternary deposits. Further support for a correlation between pedogenesis and magnetic susceptibility in the Red Clay is provided by a good correlation between magnetic susceptibility and the Rb:Sr ratio, an independent weathering index. However, differences in the magnetic mineralogy of the two units are also evident. The minimum magnetic susceptibility of the Red Clay is less than half the value of the overlying Quaternary deposits at locations such as Xifeng and Lingtai, while the maximum SP content is (relatively) larger than for the Quaternary deposits. While we cannot discount the possibility that the former difference is a reflection of different parent materials for the two units, climate may also be a contributory factor in explaining these differences. Precipitation is a major control on magnetic enhancement in paleosols [Maher and Thompson (1995) Quat. Res. 44, 383–391; Liu et al. (1995) Palaeogeogr. Palaeoclimatol. Palaeoecol. 113, 243–248], but soil anoxia and iron oxide dissolution occur if moisture exceeds some critical level. Soil moisture is therefore a critical factor in determining whether the correlation between pedogenesis and magnetic susceptibility in loess deposits is positive (central Loess Plateau, China; magnetic enhancement), negative (Alaska; Siberia; magnetic destruction) or uncertain (Pakistan; Argentina; New Zealand; cycles of enhancement and destruction). While our results indicate a generally positive correlation between pedogenesis and magnetic susceptibility in the Red Clay, they also imply that a more thorough paleoclimatic interpretation can only be achieved using complementary, but independent techniques. More work is required to deconvolve the climatic record of the Red Clay, but the reward may be the extension of the record of paleomonsoon evolution back into the Pliocene and late Miocene.