Morphological evidence suggests that GnRH may be released into cerebrospinal fluid (CSF) of the third ventricle. Therefore, a method of cannulating the third ventricle of monkey brains was developed for the purpose of examining GnRH secretion in primates. A stainless steel guide cannula was stereotaxically implanted into the third ventricle of 14 ovariectomized rhesus monkeys. A Silastic cannula for collecting CSF was inserted via the guide cannula into the ventral portion of the ventricle, permitting repeated CSF sampling for long time periods from the same animal. One week to 6 months after cannulation, CSF was collected continuously for periods of 5-10 h at 2 different rates (480 and 120 microliter/h) from conscious monkeys seated in chairs. Samples were divided into 15-min fractions, and the GnRH concentration in each was determined by RIA. In contrast to most previous studies, third ventricular CSF was found to contain significant concentrations of GnRH. GnRH was detected in 40 of 50 collections. Concentrations ranged from less than 8 to greater than 800 pg/ml, a range similar to that observed in hypophyseal portal blood. Furthermore, fluctuations within individual collections indicated that GnRH was released in pulses. The mean GnRH pulse frequency during the higher rate of CSF withdrawal was 0.43 +/- 0.06 pulses/h (n = 31), while the mean pulse amplitude was 91 +/- 7 pg/ml (n = 64). Neither parameter was influenced by the rate of CSF removal, as frequency was 0.52 +/- 0.08 pulses/h (n = 19) and amplitude was 94 +/- 11 pg/ml (n = 82) during the lower collection rate. However, the CSF withdrawal rate had a profound influence on LH secretion. In 12 of 17 collections at the higher rate, LH levels plummeted to undetectable concentrations during the first 2 h of CSF exfusion and remained low throughout the collection period. Pituitary responsiveness was not reduced, as a GnRH bolus (0.25 or 2.5 micrograms) after 6 h of CSF removal elicited a dose-dependent stimulation of LH secretion. In contrast, a higher incidence of normal pulsatile LH secretion (12 of 19 collections) was observed when the CSF withdrawal rate was reduced. During these 12 collections, LH and GnRH pulses occurred at regular intervals and exhibited similar pulse frequencies (mean +/- SE, 0.76 +/- 0.07 and 0.67 +/- 0.09 pulses/h for LH and GnRH, respectively). Most GnRH and LH pulses were synchronized, as 86% of all GnRH pulses (43 of 50) were accompanied by a LH pulse.(ABSTRACT TRUNCATED AT 400 WORDS)