The time inside the mutant selection window (T MSW ) has been shown to be less predictive of selection of fluoroquinolone-resistant bacteria than the ratio of the area under the concentration–time curve to minimum inhibitory concentration (AUC/MIC). To explore the different predictive powers of T MSW and AUC/MIC, enrichment of ciprofloxacin-resistant mutants of four Escherichia coli strains was studied in an in vitro dynamic model at widely ranging T MSW values. Each organism was exposed to twice-daily ciprofloxacin for 3 days. Peak antibiotic concentrations were simulated to be close to the MIC, between the MIC and the mutant prevention concentration (MPC), and above the MPC, with T MSW varying from 0% to 100% of the dosing interval. Amplification of resistant mutants was monitored by plating on medium with 8× MIC of the antibiotic. For each organism, T MSW plots of the area under the bacterial mutant concentration–time curve (AUBC M ) exhibited a hysteresis loop: at a given T MSW that corresponds to the points on the ascending portion of the bell-shaped AUBC M –AUC/MIC curve [when the time above the MPC (T >MPC ) was zero], the AUBC M was greater than at the same T MSW related to the descending portion (T >MPC >0). A sigmoid function fits these separate data sets well for combined data with the four organisms (r 2 =0.81 and 0.92, respectively), in contrast to fitting the whole data pool while ignoring the AUC/MIC–resistance relationship (r 2 =0.61). These data allow the appropriate use of T MSW as a predictor of bacterial resistance.