1. In anaesthetized dogs, resting mean penile artery pressure (p.a.p.) and corpus cavernosum pressure (c.c.p.) were 70-100% and 10-15% of mean systemic blood pressure, respectively. 2. Stimulation of the pelvic nerve at 10 Hz produced an immediate drop in p.a.p. and c.c.p., followed 10-30 s later by a rise in c.c.p. to the level of p.a.p. This level was 60-90% of systemic pressure, and was maintained throughout stimulation. 3. The threshold for a rise in c.c.p. was 3-5 Hz. Atropine (1 mg/kg), phentolamine (200 micrograms kg-1) and propranolol (200 micrograms kg-1) had no effect on the response to pelvic nerve stimulation. 4. C.c.p., p.a.p. and their changes in response to pelvic nerve stimulation were not significantly altered by either stimulation or section of the hypogastric nerves. 5. Cutting the sympathetic chain on both sides at L5, or administration of phentolamine, had no effect on resting c.c.p. or p.a.p. However, subsequent responses to pelvic nerve stimulation were enhanced. 6. When the pelvic nerve was stimulated during excitation of the sympathetic chain, there was still an initial drop in p.a.p. and c.c.p. but the subsequent increase in c.c.p. was delayed or abolished. These effects were mimicked by close arterial injection of phenylephrine and blocked by alpha-adrenergic antagonists. 7. This study suggests that erections in response to pelvic nerve stimulation result from an initial increase in volume of the corpus spongiosum, followed 20 s later by a stiffening of the corpus cavernosum as its pressure increases. Only the latter process is inhibited by activity of the sympathetic fibres.