Soil aggregate stability to raindrop impact is df fundamental importance to many aspects of the soil system, in particular rainsplash erosion. This stability has been measured by use of single drop rainfall simulators, the characteristics and operation of which have, in the past, shown considerable variation. The drop sizes, impact frequency, drop fall height, chemistry of drop-forming liquid, aggregate pretreatment, and definition of breakdown affect the way in which the the apparatus can be used to define stability. In this thesis an apparatus is developed that is believed to give results with increased consistency and replicability. An index of stability (F) is evolved from data collected using the raindrop simulator which is then used to characterise a set of iron rich, tropical soils from Kenya and Mo~ambique. Information about the iron, textural, chemical and magnetic status of each soil sample is also collected and used to try to predict the stability index through a process of statistical modelling. Models for highly stable and poorly stable soils are developed and finally a definitive model that can be used for all the soils. The last section of this investigation attempts to try and explain the results in terms of the processes that might be operating in the soil. Specific reference is made to the iron content of the soils because of its natural abundance in the samples considered. The possible contributions of total iron, crystalline iron, amorphous iron and organic iron to aggregate stability are discussed.