For 10 years the 700-bed Minneapolis Veterans Affairs Medical Center has conducted a policy of carefully controlled aminoglycoside usage and monitoring of resistance of over 25,000 aerobic and facultative gram-negative bacillary isolates to the aminoglycosides. On two occasions during the 1980s, our experience of introducing amikacin at a high level of usage was associated with a significant reduction in resistance to gentamicin and tobramycin among gram-negative bacilli. Rapid reintroduction of gentamicin usage in 1982 after the first amikacin period was associated with a significant and rapid increase in gentamicin and tobramycin resistance. However, in 1986, gentamicin was again reintroduced to this institution at an initially modest level, and the percentage of usage of gentamicin was gradually increased over a 15-month period without a significant change in resistance to gentamicin, tobramycin, or amikacin while maintaining an overall 68% gentamicin usage and 30% amikacin usage. Aminoglycoside usage (measured as patient days) rose steadily from under 2,000 patient days per quarter in 1980 and 1981 to over 3,000 days per quarter in 1985. Since 1985, usage has declined to under 2,500 patient days per quarter in 1990. This usage rise and fall occurred during a steadily declining daily patient census that was 590 in 1980 and 465 in 1989. A move to a new hospital building in June 1988 was associated with an additional significant decline in resistance to all aminoglycosides (P less than 0.05), continuing a trend that was evident for the year preceding the move. Resistance to aminoglycoside antibiotics is now at the lowest level in 10 years at this institution, with only one gram-negative organism, Pseudomonas aeruginosa, that exhibits more than 5% resistance to gentamicin and no gram-negative species that are more than 5% resistant to amikacin and tobramycin.