Microbial enzymes used in food processing are typically sold as enzyme preparations that contain not only a desired enzyme activity but also other metabolites of the production strain, as well as added materials such as preservatives and stabilizers. The added materials must be food grade and meet applicable regulatory standards. The purpose of this report is to present guidelines that can be used to evaluate the safety of the metabolites of the production strain that are also present in the enzyme preparation, including of course, but not limited to, the desired enzyme activity itself. This discussion builds on previously published decision tree mechanisms and includes consideration of new genetic modification technologies, for example, modifying the primary structure of enzymes to enhance specific properties that are commercially useful. The safety of the production strain remains the primary consideration in evaluating enzyme safety, in particular, the toxigenic potential of the production strain. Thoroughly characterized nonpathogenic, nontoxigenic microbial strains, particularly those with a history of safe use in food enzyme manufacture, are logical candidates for generating a safe strain lineage, through which improved strains may be derived via genetic modification by using either traditional/classical or rDNA strain improvement strategies. The elements needed to establish a safe strain lineage include thoroughly characterizing the host organism, determining the safety of all new DNA that has been introduced into the host organism, and ensuring that the procedure(s) that have been used to modify the host organism are appropriate for food use. Enzyme function may be changed by intentionally altering the amino acid sequence (e.g., protein engineering). It may be asked if such modifications might also affect the safety of an otherwise safe enzyme. We consider this question in light of what is known about the natural variation in enzyme structure and function and conclude tha t it is unlikely that changes which improve upon desired enzyme function will result in the creation of a toxic protein. It is prudent to assess such very small theoretical risks by conducting limited toxicological tests on engineered enzymes. The centerpiece of this report is a decision tree mechanism that updates previous enzyme safety evaluation mechanisms to accommodate advances in enzymology. We have concluded that separate mutagenicity testing is not needed if this decision tree is used to evaluate enzyme safety. Under the criteria of the decision tree, no new food enzyme can enter the market without critical evaluation of its safety.