Whether or not resistant mutants will be present before the start of antibiotic treatment of an initially susceptible population of bacteria depends on the size of the infecting population, the rate of mutation to resistance, and the amount of time that the population has been maintained. In the present investigation, we argue that for the treatment of chronic infections caused by hypermutable Pseudomonas aeruginosa of the sort frequently found in cystic fibrosis patients, mutants resistant to all single antipseudomonal drugs will almost invariably be present in a high proportion at the onset of treatment, and consequently, these strains should be considered resistant to all agents when they are used as monotherapy. Using a construct of P. aeruginosa strain PAO1 with a mutS deletion (strain PAODeltamutS), we show that when in vitro populations of less than 5 x 10(4) seemingly susceptible hypermutable bacteria are confronted with any of 11 antipseudomonal agents, mutants for which the MICs and the minimum bactericidal concentrations are in the range of clinical resistance will almost invariably ascend to dominance within 24 to 36 h. This does not occur for PAO1 without the mutS deletion. The results of our detailed analysis of this evolution of acquired resistance to two of these antibiotics, imipenem and ciprofloxacin, indicate that although the rates of mutation to resistance in PAODeltamutS are on the order of 1 x 10(-6) per generation, resistant mutants are very likely to either be present in cultures of between 2 x 10(4) and 4 x 10(4) bacteria or arise after the bacterial populations are confronted with antibiotics. We also demonstrate with in vitro experiments that the problem of acquired resistance to treatment with single antibiotics can be thwarted by combination therapy with pairs of antibiotics of different classes with synergistic activities. We discuss the clinical implications of our analysis of these observations.