SYNCHEMB is a large knowledge-based domain-specific heuristic problem solving system that applies the methodology of artificial intelligence to the problem domain of synthetic organic chemistry. The program may be used for both synthetic and retrosynthetic exploration of the reaction transform space. We describe some experiments that might lead to the development of a special purpose version of the program for the investigation and prediction of environmental and metabolic processes undergone by chemicals that are released into the biosphere or directly ingested. The test system was used to predict the metabolites of three molecules of considerable biochemical interest, benzo[alpyrene (known to be metabolically transformed into an active carcinogen), paramethyl substituted phenylalanine (which undergoes the NIH-shift when metabolized), and haloperidol (an important and widely used neuroleptic), In the case of the benzo[alpyrene, the program predicted all of the known metabolites, as well as several others. All of the 11 compounds predicted as metabolites of 4-methylphenylalanine, which included the major known metabolite that exhibits the NIH-shift, were reasonable. Contained among the predicted metabolic pathways for haloperidol was a credible explanation for the occurrence of a known haloperidol metabolite derived from a substituted phenylacetic acid, but for which no intermediates have been isolated. The program was able to bridge the problem space between haloperidol and the derived hippuric acid by using an unsuspected, but metabolically sound pinacol rearrangement.