The stability of the erosion system in a tropical rain forest environment is investigated using the model previously described by Morgan, Morgan and Finney (1982) and validated by Morgan and Finney (1982). Simulations are carried out for the natural primary forest cover, commercial timber extraction and the agricultural ecosystems of rubber cultivation, shifting cultivation with sixteen-year, fourteen-year and four-year cycles, and continuous rotational cropping of groundnuts, maize and potatoes. The model is operated for five slope steepnesses using a hundred-year synthetic sequence of rainfall data generated by Monte Carlo analysis. Erosional stability is expressed in terms of changes in rooting or top soil depth. Using the model it is possible to distinguish between three broad states of biostasy, rhexistasy and homeostasy, representing increasing, decreasing and stable rooting depths respectively. Detailed examination of the magnitudes of the change in rooting depth in response to perturbations in the system and the length of time required for recovery can show the extent to which systems in a state of or close to homeostasy are potentially unstable.