Particulate organic matter (POM) is a major vector for carbon and nutrients in stream ecosystems. The decomposition process by which this material enters the food web has been difficult to study directly because in-situ degradation rates for POM are difficult to obtain, particularly for fine particles. We evaluated a new statistical method that estimates POM mass loss rates from indices of microbial enzyme activity. Litter bags containing benthic POM in three size ranges--Fine (F) 0.063-0.25, Medium (M) 0.25-1 and Coarse (C) 1-4 mm--were placed in a eutrophic woodland stream in June 1992. Over a 6-mo period, samples were collected and analyzed for mass loss and the activities of seven extracellular enzymes involved in the degradation of holocellulose, chitin, and polyphenolic compounds. In-situ POM samples were collected concurrently and assayed for the same suite of enzyme activities. In all cases, mass loss was related to cumulative enzyme activity; however, apparent enzymatic degradation efficiencies were generally lower for MPOM and FPOM (=MFPOM) than for CPOM by factors of 1.5-7. Separate regression models, relating mass loss to average cumulative enzyme activity, were developed for CPOM and MFPOM and used to estimate in-situ decomposition rates. Mean model-estimated turnover times for in-situ CPOM, MPOM, and FPOM were 127, 145, and 185 d, respectively, compared with values of 143, 312, and 244 d determined for size-sorted POM confined in litter bags. The application of enzymatic decomposition models may circumvent some of the methodological problems that have limited the study of POM dynamics in lotic ecosystems.