This paper describes the outline of an approximate mathematical technique capable of including stochastic components in environmental models based on ordinary differential equations. The technique is based on the decomposition of the solution into additive components, with the first component being the solution of a simplified deterministic equation and each of the other components being found in terms of those preceding it. Moments of the solution process can then be calculated by performing appropriate averaging. The method provides natural statistical separability, that is, no truncations in the derivation stage or closure approximations are necessary. The random quantities involved can be either random variables or stochastic processes. No unrealistic assumptions, like white noise behavior or small randomness, are necessary. Solutions, which are in the form of infinite but convergent series, can be easily implemented into computer programs, providing convenient support in management and decision analysis. A demonstration application of a simple biochemical oxygen demand model for river waters illustrates the usefulness of the technique in environmental modeling. ¿ American Geophysical Union 1993