Antibiotic resistance is generally thought to arise from mutations in pre-existing genes or the acquisition of resistance-conferring genes from other microbes. However, when a population of Pseudomonas aeruginosa is exposed a second time to the antimicrobial agent gentamicin, it displays a temporary, but higher resistance than it did in the first exposure. The cells return to control susceptibility levels again after approximately six hours growth in the absence of selection. This phenomenon has been called adaptive resistance. The central question addressed by this study is whether this transient resistance pattern is due to a change in the physiology of individual cells following first exposure to the drug, or to the selection of pre-existing resistant mutants that are uncompetitive in the absence of the drug. In contrast with previous studies, where the transient resistance pattern was induced with relatively high concentrations of gentamicin, this study found that pre-exposing cells to sub­ inhibitory concentrations of gentamicin were most effective at reproducing the adaptive resistance phenotype. The origin of the resistance phenotype was investigated using fluctuation analysis. If the adaptive resistance phenotype is within the capacity of all the cells in a bacterial population then we would predicted the number of resistant bacteria to cluster around a mean. Instead, different cultures held significantly different numbers of resistant bacteria, a result consistent with a mutational origin of gentamicin resistance. Sib selection (a procedure that allows for the selective enrichment of resistant bacteria without exposure to the selective agent) was used to isolate putative pre-existing mutants. In contrast to the results of the fluctuation analysis, a pure culture of resistant bacteria could not be established. Competition experiments between gentamicin resistant and sensitive cells confirmed that at least some resistance phenotypes result in slower growth in the absence of the drug.