The prediction of urine antibacterial activity from pharmacological and microbiological parameters was assessed by using experimental urine levels and urine bactericidal titers determined up to 72 h after a 400-mg single dose of two quinolones in a phase I study. The area under the bactericidal curve (AUBC) was accurately predicted for norfloxacin but significantly (P < 0.001) underestimated for rufloxacin (actual value was four times higher than the predicted value against Escherichia coli and two times higher against Staphylococcus aureus). In vitro susceptibility differences between the two strains predicted the ex vivo AUBC differences for norfloxacin but not for rufloxacin, where ex vivo differences were greater than expected. Urine bactericidal titers for up to 72 h were accurately predicted for norfloxacin against E. coli and S. aureus and for rufloxacin against S. aureus, but experimental activity for up to 48 h was four times higher (P < 0.001) than the predicted activity for rufloxacin against E. coli. In the case of norfloxacin, the duration of adequate urine antibacterial activity against S. aureus was overestimated. Inaccurate estimations of ex vivo antibacterial activity of a suspected active metabolite (as with rufloxacin) when an adequate cutoff is not established may have dosing implications.