A series of experiments was conducted to ascertain the significance of 'small' pulses of gonadotropin-releasing hormone (GnRH). In the first experiment, ovariectomized hypothalamo-pituitary disconnected (HPD) ewes were given 250 ng pulses of GnRH every 2 h for 1 week, 25 ng pulses every 2 h for 24 h, 25 ng pulses hourly for 24 h and then alternating hourly pulses of 25 and 250 ng. During the 25 ng pulses, LH was not detectable in plasma and FSH concentrations declined after 2 days. Following the 25 ng pulses, the resumption of 250 ng pulses led to exaggerated LH responses (mean ± S.E.M. pulse amplitude 18.7±1.7 vs 10.2±1.2 μg/l in the first week). In a second experiment, ovariectomized-HPD ewes were maintained on 250 ng GnRH pulses every 2 h for 1 week and were then given three 25 ng pulses mid-way between the 250ng pulses. Samples of blood were taken over three 250 ng pulses without 25 ng insertions and over three pulses with insertions. The insertion of 25 ng GnRH pulses did not cause LH pulses in their own right and did not alter the LH responses to the 250 ng pulses. In a third experiment, 50 ng GnRH pulses were inserted between the 250 ng GnRH pulses, as in experiment 2; these 50 ng pulses caused small LH pulses and led to a reduction in the response of the LH pulse amplitude to the 250 ng pulses. The 'small' LH pulses which occurred in response to 50 ng GnRH compensated for the reduced responses to the 250 ng pulses. Hence, the integrated area under the LH curve and between successive 250 ng pulses remained the same, irrespective of the 50 ng insertions. From these data we conclude that 'small' GnRH pulses alone can sustain ongoing LH synthesis without release, leading to an accumulation of releasable LH, and that the insertion of 'small' GnRH pulses may modify the pattern of pituitary responsiveness to 'large' GnRH pulses.