In order to delineate factors contributing to variation in hormone levels, progesterone and prolactin (PRL) levels from 28 normal women, obtained daily during one menstrual cycle and every 20 minutes during a midluteal 24-hour admission in a subgroup of five subjects, provided a data base for analysis of these variables. Pulsatile analysis of the 24-hour data was conducted using an adaptive-threshold algorithm, and normal reference ranges were generated from randomly selected daily hormone values. Our data verify that inherent variation can significantly alter single random serum levels of reproductive hormones. These variations included menstrual cycle day, circadian influence, pulsatile secretion, assay error, and biologic heterogeneity. Besides the expected day-to-day change in progesterone levels during the luteal phase, seven of ten women exhibited a significant circadian variation in progesterone; however, the time of day of the peak level was not consistent among women. Prolactin levels did not demonstrate any clinically relevant change over the menstrual cycle, but did have a consistent circadian pattern (nocturnal rise) over the 24-hour study period. Pulsatile variation occurred in both progesterone and PRL levels during the 24-hour admission. Five different reference ranges were generated from randomly selected single daily values from the 28 normal menstrual cycles. Although the mean levels calculated for each reference range were similar, the reference ranges demonstrated considerable variation due to the random sampling. In the five progesterone reference ranges, the lower limit of the range varied from 2.7-6.1 ng/mL, whereas the upper limit varied from 24.2-42.1 ng/mL.(ABSTRACT TRUNCATED AT 250 WORDS)