Lungs from near-term fetal guinea pigs were supported in vitro for 3 h; lung liquid production rates were measured by a dye dilution technique. Seventy preparations were used to study the effects of arginine vasopressin (AVP) placed in the outer saline for the middle hour, at concentrations reported at birth [fetuses 61 +/- 2 days of gestation; 94.7 +/- 16.2 g (SD) body weight]. At 1200 microU/ml, AVP arrested fluid production (rates, successive hours, 3.03 +/- 0.60, 0.50 +/- 0.14 and 0.02 +/- 0.08 ml/kg body weight per h; falls significant, P < 0.01-0.0005). At 600, 300 and 100 microU/ml there were significant but smaller reductions. Reabsorptions were seen in 8 preparations given 600-1200 microU/ml, AVP. Preparations given 10 microU/ml AVP, AVP carrier or control saline showed no significant change. The responses (% reductions during treatment), were linearly related to the log concentration of AVP (r = 0.99); theoretical threshold, 8 microU/ml). Increasing treatment to 2h did not increase final responses. Preparations from 5 fetuses > 120 g body weight showed significantly greater responses (P < 0.025) [fetuses 64 +/- 2 days of gestation; 135.1 +/- 18.6 g (SD) body weight]. 10(-6) M amiloride abolished responses to AVP [fetuses 62 +/- 1 days of gestation; 93.4 +/- 18.5 g (SD) body weight, n = 30; rates, succeeding hours; AVP alone, 1.78 +/- 0.22, 0.48 +/- 0.09, 0.16 +/- 0.99 (P < 0.01-0.0005); AVP with amiloride, 1.15 +/- 0.07, 0.93 +/- 0.10, 0.86 +/- 0.08 (no significant fall) ml/kg body weight per h]. Thirty-six preparations treated with arginine vasotocin (AVT, 10-600 microU/ml) showed closely similar responses to those from AVP. These studies extend results to fetal guinea pigs, and show that AVP, at concentrations reported at delivery, can slow lung liquid production or cause reabsorption by a direct action on the lung. The effect increases close to term, and is due to activation of amiloride-sensitive Na+ channels.