Previous attempts to characterize the estrous cycle of elephants have yielded conflicting estimates of cycle length and LH profiles. In order to establish artificial breeding programs in this species, resolution of these issues is needed. Therefore, four female African elephants housed at the Indianapolis Zoo were studied for approximately 6 mo beginning in December 1994. Blood was collected weekly, and the serum was immediately analyzed for progesterone (P4). Whenever the weekly concentration of P4 was found to be low, blood was collected one or four times per day. All serum samples were assayed for LH, and the daily samples were assayed for P4 and estradiol. Transient increases of serum LH (designated as peaks) were observed four times in each of the four females. Of these 16 LH peaks, 8 were classified as ovulatory LH (ovLH) peaks and 8 were classified as anovulatory LH (anLH) peaks. Peaks designated ovLH averaged 3.60 +/- 0.67 ng/ml (mean +/- SEM); serum P4 measured during these peaks began to increase 2-3 days before each ovLH peak and continued to increase for several weeks thereafter, reaching a peak of 675 +/- 35 pg/ml. The eight other LH peaks, designated anLH peaks, were of similar (p > 0.05) magnitude averaging 3.07 +/- 0.72 ng/ml, but the serum concentration of P4 remained very low (< 80 pg/ml) during and for several weeks after these peaks. Six peaks designated anLH occurred an average of 12.2 +/- 1.4 days after serum P4 had declined below 80 pg/ml. In each elephant, there was a regular sequence in which each ovLH peak was followed by a luteal-active period lasting about 60 days and then about 12 days later by one anLH peak. Each anLH peak was followed 19-22 days later by one ovLH peak, but serum P4 remained at nonluteal levels throughout this interval between peaks. The authors propose to designate this interval after the anLH peak and before the next ovLH peak as a nonluteal (i.e., low P4) estrous cycle of only 3-wk duration. Following each short nonluteal estrous cycle, there was a single ovLH peak that initiated one luteal-active estrous cycle lasting 10-11 wk until terminated by the next anLH peak. The present results demonstrate that nonpregnant African elephants, housed in the absence of males, alternate between short nonluteal estrous cycles and long luteal-active estrous cycles. Daily measurements of serum P4 can be used to distinguish between the two types of estrous cycles and thereby provide a clinical prediction about the optimum time for artificial insemination.