The time inside the mutant selection window (MSW), T(MSW), appears to be less predictive of the selection of fluoroquinolone-resistant Staphylococcus aureus than is the ratio of the area under the concentration-time curve (AUC) to the MIC. This observation might be attributed to the fact that T(MSW) does not consider the actual position of simulated antibiotic concentrations inside the MSW, which also might influence the amplification of resistant mutants. To test this hypothesis, the enrichment of ciprofloxacin-resistant S. aureus was studied at ciprofloxacin (CIP) concentrations that oscillate near the mutant prevention concentration (MPC), i.e., closer to the top of the MSW ("upper case"), and closer to the MIC, i.e., at the lower limit of the MSW ("lower case") at the same T(MSW). Two methicillin-resistant strains of S. aureus, ATCC 6538 and ATCC 43300 (MICs of 0.25 and 0.5 mg/liter, respectively, and MPCs of 4 and 2 mg/liter, respectively), were exposed to twice-daily CIP treatments for three consecutive days. With S. aureus ATCC 6538, the simulated ratios of the AUC at 24 h (AUC(24)) to the MIC were 50 and 260 h (T(MSW) 75% of the dosing interval). With S. aureus ATCC 43300, the simulated AUC(24)/MICs were 30 and 100 h (T(MSW) 56%). With each organism, mutants resistant to CIP were enriched in an AUC(24)/MIC-dependent manner: the higher the AUC(24)/MIC ratio, the lower the growth on CIP-containing plates. For example, the area under the time-kill curve of mutants resistant to 4x MIC of CIP in the upper case was three times smaller than that in the lower case for both S. aureus strains. Similar differences were seen at the higher (8x MIC) and lower (2x MIC) CIP concentrations. These data highlight differences in the selection of resistant S. aureus, depending on the position of simulated concentrations inside the MSW at a given T(MSW). This explains why T(MSW)-based predictions of resistance are less accurate than those based on AUC/MIC and AUC/MPC.