Fitness represents the combined effects of all other phenotypic properties on the capacity for survival and reproduction by a particular genotype in a particular environment. For most environmental application of genetically modified microorganisms, efficacy will be enhanced if the engineered genotype is more fit than its wild-type counterpart in the target environment. However, inadvertent spread of the engineered genotype will be less likely if it is less fit than the wild-type. Thus, the fate of a population of genetically engineered microorganisms, and the likelihood and magnitude of any environmental effects (whether beneficial or detrimental), will be strongly influenced by the relative fitnesses of modified and unmodified genotypes. In this chapter, I have presented theoretical principles and empirical methods for determining the relative fitnesses of engineered and wild-type clones. Selection coefficients were used to provide a quantitative measure of the difference in fitness between the two clones in a particular environment. Many engineered genotypes are unstable, such that their frequencies decline with time. Instability may be caused by infidelity of replication or transmission of a particular gene (which is termed segregation), or it may be caused by a difference in the fitness of genotypes that retain or have lost that gene (selection). In this chapter, I have also presented theoretical principles and empirical methods for distinguishing the effects of selection and segregation. Finally, it should be emphasized that selection coefficients and segregation rates can be estimated not only in highly simplified laboratory systems, but also in more complex natural or semi-natural systems, such as microcosms. All that is required is the ability to monitor the relative abundance of two clones (e.g., engineered and wild-type) that share a common environment.