The objective of this study was to find the earliest time at which it was possible to detect clinical pregnancy in an invitro fertilization (IVF) treatment cycle supported with human chorionic gonadotrophin (HCG), and also retrospectively to diagnose abnormal ovarian-or endometriumrelated situations in failure cycles. Serum samples were taken in 41 IVF cycles at frequent intervals from the beginning of ovarian stimulation until menstrual bleeding occurred or a pregnancy was established. Concentrations of oestradiol, progesterone, placental protein 14 (PP14), pregnancy-specific pi-glycoprotein (SPj), and pregnancyassociated plasma protein A (PAPP-A) were determined in the serum samples using commercially available (steroid) or purpose-developed (protein) immunoassays. The cycles were retrospectively distributed into four outcome groups: (i) fertilization failure (FF, n = 8); (ii) implantation failure (IF, n = 10); (iii) 'interaction' (embryo-endometrium) cycle (IC, n = 14), and (iv) clinical pregnancy (CP, n = 9). The embryo-endometrium interaction was detected by a rise in SP! in 23 cycles (70% of embryo transfers) at a time when endogenous HCG was still masked by external support. Early ('false') positive SP] concentrations were observed in two out of eight and five out of 14 cases in groups FF and IC respectively, but never amongst the ongoing pregnancies (CP). PAPP-A did not distinguish pregnancy from the other outcomes. The PP14/progesterone ratio was lower, later in the cycle, in CP than in the other groups. We conclude that, while it is not possible to predict the outcome of a given IVF cycle earlier than 2 weeks after embryo transfer, the hormonal patterns can be used to detect abnormalities (e.g. endometrial asynchrony) which may be useful for subsequent treatment cycles in the same patient.