The stability of soil aggregates against water erosion is a largely unstudied factor in research on the spatially-distributed impacts of tropical forestry. Soil strata with particularly weak aggregates are likely to be the focus for erosional activity whether buried or exposed by either natural processes or forestry activities. Understanding the location and likely cause of such intrinsic instability would, therefore, allow better spatially-distributed parameterisation of erosion models. This study has aimed to identify the range in aggregate stability within a region of Bornean Ultisol disturbed by forestry operations, and to identify the bulk soil properties associated with such stability variations. The sites sampled were subject to a range of denudational processes including piping, rilling and landslide-triggered erosion. Soil profiles with rates of erosion in excess of 10 mm a−1 were shown to have less aggregate stability (as characterised by the rainfall simulation survival index (RSSI)) than those with no visible signs of erosion. Further, large differences in aggregate stability between soil horizons of the same profile were observed and seen to be statistically correlated with the bulk soil properties of organic carbon and clay content (i.e. stabilising agents) and the dispersing agent of exchangeable sodium percentage (ESP) at sites undergoing erosion. Organic carbon appeared to be the most important governing factor, accounting for 56% of the variance in the aggregate stability. The presence of strongly and weakly expanding 2 : 1 clays within the soils may have been an additional de-stabilising factor. Reduced porosity was also observed at eroding sites with weak soil aggregates. The properties of EC25 and sodium adsorption ratio (SAR) were not correlated with the RSSI or the presence of soil erosion. Identification of the role of organic carbon, clay and ESP in the stability of these tropical forest aggregates is important in focusing future, more intensive research on the spatial parameterisation of models to simulate forestry impacts on erosion.