To ascertain whether monitoring of the concentrations of ofloxacin in saliva during a course of treatment is more suitable and safer than that of its levels in blood, we simultaneously monitored its concentrations in three body fluids (blood, saliva, and expectorated sputum) after a 300-mg administration in 18 patients with chronic respiratory infection. The mean (+/- standard error of the mean) half-lives derived from the three drug level-time relationships were similar: 6.04 +/- 0.58 h for serum, 6.34 +/- 0.63 h for sputum, and 6.61 +/- 0.65 h for saliva. The mean peak concentration (4.06 to 4.53 micrograms/ml) did not differ at the three sites, but the times taken to reach peak concentration in saliva and sputum (3.17 +/- 0.46 h) were significantly longer than that in serum (2.22 +/- 0.28 h). The ratios of the concentrations in saliva and sputum to the concentration in serum increased during the first 2 h and reached 1.0 between 2 and 8 h after administration. They rose above 1.0 16 h after administration: 1.14 +/- 0.11 for saliva and 1.19 +/- 0.10 for sputum. The concentration-time relationship for sputum corresponded closely with the concentration-time relationship for saliva, and an overall significant correlation between the concentrations in sputum and saliva was obtained (P < 0.01). These results suggest that monitoring concentrations in saliva may be more valid, as well as less invasive, than monitoring of the levels in blood for ensuring that the drug concentration reaches its therapeutic level in bronchial secretions.