On two occasions at least 1 week apart, nine healthy male volunteers were administered in random order either 2 x 200 mg (1.8 mmol) acyclovir tablets or 400 mg of acyclovir in 500 ml of 5% dextrose solution which was infused at constant rate into the duodenum over 4 h. Six of the subjects subsequently sipped the same solution at the rate of 10.4 ml in each 5 min period for 4 h. Blood and urine were sampled over 24 h for each mode of administration. Acyclovir was assayed by radioimmunoassay. Mean areas under the plasma concentrationtime curves (AUCs) + s.d. for tablet (T), intraduodenal infusion (I) and sipping (S) were, respectively: T = 14.7 + 5.1; I = 24.6 + 5.1; S = 28.4 + 9.5 (n = 6) ,umol 1-1 h. AUCs for I and S were significantly greater than that for T (2P < 0.05). Mean apparent maximum plasma concentrations (Cmax) + s.d. were T = 3.8 + 1.5; I = 4.8 + 0.9; S = 5.1 + 1.5 ,umol 1-1. This trend to higher values for I and S was not significant. Mean apparent plasma disappearance half-lives (ttl2) + s.d. were respectively T = 2.3 + 0.4; I = 2.7 + 0.5; S = 3.0 ± 0.2 h, I being significantly greater than T (2P < 0.05), as was S greater than T (2P < 0.01). The mean 24 h urinary recovery + s.d. of acyclovir (n = 8) was T = 309 + 119 ,umol (17% of dose); I = 471 + 83 ,imol (26% of dose); S = 507 ± 180 ,umol (28% of dose). I > T (P < 0.01), S > T (P < 0.05). Acyclovir absorption is increased when contact time of a solution with the absorptive area of the human gut is increased as compared with absorption available from the tablet. This suggests capacity limited absorption.