Antibiotic consumption during 1996 was measured in 15 large hospitals from 14 countries and 3000 consecutive Staphylococcus aureus samples were collected, allowing calculation of local resistance rates and typing of isolates. Antibiotic consumption data were converted to defined daily doses (DDD), and similar antibiotics were grouped if they belonged to the same therapeutic subgroup. Variations in hospital size were corrected by using DDD per 1000 bed-days. The total antibiotic consumption in the 15 hospitals varied between 296 DDD/1000 bed-days and 1108 DDD/1000 bed-days. Differences in the usage of therapeutical subgroups of antimicrobials varied significantly between hospitals. A positive correlation was found between S. aureus resistance to methicillin (MRSA) and consumption of beta-lactam combinations, between resistance to quinolones and consumption of beta-lactam combinations and carbapenems and resistance to aminoglycosides and consumption of beta-lactam combinations. The consumption of beta-lactamase-sensitive antibiotics was negatively correlated to resistance to methicillin, quinolones, and aminoglycosides. Usage of the different antimicrobial therapeutical subgroups was also correlated. Consumption of beta-lactamase-sensitive antibiotics (penicillin) was positively correlated to consumption of beta-lactamase-resistant penicillins and negatively correlated to consumption of carbapenems, quinolones, and glycopeptides, whereas consumption of cephalosporins was positively correlated to consumption of aminoglycosides, quinolones, and glycopeptides. In this study of hospitals with MRSA prevalence of between 0% and 63%, significant correlations were found between resistance and consumption of antimicrobials. These findings support the importance of antimicrobial consumption on resistance. An accompanying paper addresses the issue of antibiotic resistance and clonality of isolates.