Time series analysis was used to detect LH and FSH episodes in untreated seasonally anoestrous ewes and prepubertal heifers, and in these animals when treated with low doses of GnRH. For comparison, these profiles were also assessed for episodic secretion by subjective, visual appraisal methods and by cycle detection-an objective threshold method. In untreated animals, time series analysis detected recurring events in the LH and FSH profiles, the period lengths of which varied between individual animals. When GnRH was injected at 2-h intervals, cycles in LH secretion with period lengths of 120 min were recorded in all animals, of 60 min in all ewes and 11/12 heifers, and of 40.5 min in 22/24 ewes and 10/12 heifers. The cycles with period lengths of 60 and 40.5 min are probably artefacts of this method of analysis. No consistent cycles in FSH release were detected in GnRH-injected anoestrous ewes, but 120-min cycles were recorded in 8/12 GnRH-injected heifers. When GnRH was administered to seasonally anoestrous ewes by continuous infusion, recurring cycles in both LH and FSH secretion were evident. However, there was no consistency in their period lengths and the mean number and frequency of cycles were similar to pretreatment values. The number of episodes detected by visual appraisal was influenced by the choice of episode definition. Both methods identified LH, but not FSH, episodes in response to each injection in all GnRH-injected animals. Cycle detection, which does not identify individual episodes, recorded LH and FSH episode frequencies similar to those detected by the more stringent method of visual appraisal. Time series analysis detected an FSH response to GnRH injections in prepubertal heifers that was not identified by the other methods of analysis. However, because of the asymmetric nature of LH episodes, it also detected cycles in LH profiles that were probably spurious. Subjective decisions influenced the frequencies of LH and FSH episodes recorded by visual appraisal, and the variation in episode amplitude in these profiles made cycle detection inappropriate. Each of these methods can contribute to the interpretation of hormone profiles, but their constraints and limitations must be recognized.