When toadfish are made ureotelic by a crowding/confinement protocol, they excrete approximately 90 % of their urea nitrogen (urea-N) production in large, irregular pulses (1-2 pulses per day) from the gill region. We investigated three hypotheses as to the mechanism of pulsatile excretion: (i) the presence of an active reabsorptive 'back-transport' mechanism that is periodically inhibited to allow urea-N excretion to occur; (ii) the periodic occurrence of a generalized, non-specific increase in gill permeability; and (iii) the presence of a specific facilitated diffusion transport system that is periodically activated. Exposure of toadfish during non-pulse periods to treatments designed to block a 'back-transport' mechanism (Na+-free sea water or the urea analogues 30 mmol l-1 thiourea or 30 mmol l-1 acetamide in the external water) did not stimulate a leakage of urea-N, thereby opposing the first hypothesis. The second hypothesis was opposed by several results. Neither injection of the potent branchial vasodilator L-isoprenaline (10(-5) mol l-1) nor infusion of NH4Cl, the latter at levels known to stimulate urea-N efflux in perfused gills, had any effect on urea-N excretion. Furthermore, during natural pulse events, when the normally very low gill permeability to urea (3x10(-7) cm s-1) increased at least 35-fold, there was no accompanying increase in permeability to either 3H2O (1.5x10(-5) cm s-1) or the paracellular marker [14C]PEG-4000 (10(-8) cm s-1). However [14C]thiourea permeability (1.5x10(-7) cm s-1) increased approximately fivefold, in support of the third hypothesis. Furthermore, when 30 mmol l-1 urea was placed in the external water, a concentration (60 000 micromol-N l-1) approximately three times that of blood (20 000 micromol-N l-1), each efflux pulse event (measured with [14C]urea) was accompanied by a net uptake, such that blood urea-N levels rose rather than fell. A proportional 1:1 relationship between influx per unit external concentration and efflux per unit internal (i.e. plasma) concentration indicated a fully bidirectional transport system. The simultaneous presence of 60 mmol l-1 thiourea in the external water inhibited the influx component by 73 %, further supporting this conclusion. These data, together with recent molecular, morphological and endocrinological evidence, strongly suggest that pulsatile urea-N excretion is caused by the periodic activation of a facilitated urea transporter in the gills, similar to the vasopressin-regulated urea transporter in the mammalian kidney.