Mathematical models currently used to study the relationship between the prevalence of the resistance to an antibiotic and the amount of drug may not be adequate because they do not integrate temporal and area analyses simultaneously. Furthermore, the forms of such relationship are unknown. We applied the Generalized Additive Model (GAM) to study fluoroquinolone use and the incidence of fluoroquinolone-resistance in Staphylococcus aureus in our institution over a 3-year period. Overall institution and individual hospital unitrestricted analyses were performed. Relative risk (RR) observed for increasing fluoroquinolone use with a 4-month lag from the 25th percentile to the 75th percentile was 1.27 (95% CI: 1.13-1.42). In the surgery units, RR was 1.23 (95% CI: 1.01-1.50) for fluoroquinolone use with a 2-months lag. GAM enabled us to observe for the first time a significant temporal relationship between fluoroquinolone use and the incidence of fluoroquinolone-resistant nosocomial S. aureus isolates.