Objective: The purpose of this study was to compare the patterns and timing of the increases in plasma levels of corticotropin-releasing hormone and unconjugated estriol during human pregnancy. Study Design: Corticotropin-releasing hormone and unconjugated estriol were measured in serial samples that were collected from preterm subjects and from spontaneous term control subjects who were selected randomly from a study cohort of 297 women. Results: Gestational increases in log corticotropin-releasing hormone and log unconjugated estriol concentrations were best described by linear and cubic polynomial functions, respectively. Plasma unconjugated estriol levels were similar in preterm and term singleton pregnancies at equivalent gestation, whereas corticotropin-releasing hormone was elevated earlier in premature subjects. Mean corticotropin-releasing hormone levels relative to term control subjects (n = 40 women) were shifted forward 16.5 ± 8.1 days (P =.027, singleton preterm, spontaneous labor; n = 16) and 33.0 ± 7.9 days (P <.001, singleton preterm, obstetric intervention; n = 10). The corresponding shifts in unconjugated estriol values were –3.8 ± 2.0 days and –2.7 ± 5.6 days (both not significant). The prematurity of delivery showed a significant regression on shifts in corticotropin-releasing hormone (P =.004 and P <.001) but not in unconjugated estriol for the 2 groups. The ability to predict prematurity was not significantly improved by regression on corticotropin-releasing hormone and unconjugated estriol shift values together. Conclusion: The patterns and timing of gestational changes in corticotropin-releasing hormone and unconjugated estriol differ in humans. The usefulness of corticotropin-releasing hormone as a biochemical preterm marker in singleton pregnancies is not enhanced by the additional measurement of plasma unconjugated estriol. (Am J Obstet Gynecol 2002;186:94-9.)