Hormone assays are being developed in the laboratory to detect specific molecular markers in nonclinical populations. Epidemiology is increasingly using these assays to improve the precision with which disease processes and exposures can be defined. This growing body of molecular epidemiology requires a high degree of cooperation between the assay developer and the assay user. We draw on our experience in using a sensitive hormone assay for the detection of early pregnancy via urinary human chorionic gonadotropin to illustrate these points. We conclude that this collaborative effort, in addition to making this study possible, has provided unexpected rewards.