The mutant selection window is defined by two boundaries. The lower boundary of the window is the lowest concentration that blocks the growth of the majority of drug-susceptible cells, since below that concentration mutant cells do not have a growth advantage. The lower boundary can be approximated by the minimal inhibitory concentration for half the cells in the population (MIC₍₅₀₎); however, inhibition of 99% of the cells (MIC₍₉₉₎) is a more suitable boundary since it is more accurately measured. The standard MIC does not approximate the lower boundary as well because some selective pressure is exerted when so many cells (10⁴ to 10⁵) are used in the measurement. Indeed, resistant mutants can be enriched by repeated passage of cells at concentrations just below MIC (6). Nevertheless, MIC must be near the bottom of the window because treatment of Staphylococcus aureus with moxifloxacin in a dynamic model shows mutant enrichment occurring above the MIC, not below it (12).

The upper limit of the window is the drug concentration that blocks the growth of the least susceptible, single-step mutant. Above this concentration, cell growth requires the presence of two or more resistance mutations. Since two concurrent mutations are expected to arise rarely, few mutants will be selectively amplified when a susceptible population is exposed to drug concentrations that exceed the upper boundary. As pointed out above, with fluoroquinolones the mutation frequency for resistance due to target (topoisomerase) mutations can be less than 10^-7; consequently, more than 10¹⁴ bacteria would be required to find a cell with two concurrent, independent fluoroquinolone-resistant target mutations. In clinical cases, bacterial populations may reach 10¹⁰ cells within an infected individual, but 10¹⁴is unlikely. Thus, resistance is expected to develop rarely when drug concentrations are kept above the upper boundary of the mutant selection window. This expectation led to the upper boundary being designated as the mutant prevention concentration (MPC).

The effect of a methoxy group is evident: it lowers the MPC by about 10-fold with M. bovis BCG. It also lowers MIC, but to a lesser extent (9).