In this review, we consider the ways in which vasopressin and oxytocin have been measured since their first discovery. Two different ways of measuring oxytocin in widespread use currently give values in human plasma that differ by two orders of magnitude, and the values measured by these two methods in the same samples show no correlation. The notion that we should accept this seems absurd. Either one (or both) methods is not measuring oxytocin, or, by "oxytocin", the scientists that use these different methods mean something very different. If these communities are to talk to each other, it seems important to validate one method and invalidate the other, or else to establish exactly what each community understands by "oxytocin". A similar issue concerns vasopressin: again, different ways of measuring vasopressin give values in human plasma that differ by two orders of magnitude, and it seems that the same explanation for discrepant oxytocin measurements applies to discrepant vasopressin measurements. The first assays for oxytocin and vasopressin measured biological activity directly. When immunoassays were introduced, they encountered problems: high molecular weight factors in raw plasma interfered with the binding of antibodies to the hormones leading to high and erroneous readings. When these interfering factors were removed by extraction of plasma samples, immunoassays gave measurements consistent with bioassays, with measures of turnover and with the sensitivity of target tissues to exogenous hormone. However, many recent papers use an enzyme linked immunoassay to measure plasma levels without extracting the samples. Like the first radioimmunassays of unextracted plasma, this generates impossibly high and wholly erroneous measurements. This article is protected by copyright. All rights reserved.