Purpose To investigate hCG-β level on postovulatory day (POD) 12 and its fold increase as predictors for pregnancy outcome after in vitro fertilization (IVF) cycles. Methods A retrospective cohort study was performed in total 1408 fresh and 598 frozen cycles between November 2008 and October 2011, which resulted in biochemical pregnancy, early pregnancy loss, or live birth of singleton pregnancy. The serum hCG-β levels of POD 12 and 14 were compared among biochemical pregnancy, early pregnancy loss, and live birth groups. The cutoff values of POD 12 and 14 hCG-β levels and the degree of hCG-β increase from POD 12 to 14 were determined for each pregnancy outcome. Results POD 12 and 14 hCG-β levels stratified based on pregnancy outcomes were significantly different among the biochemical pregnancy, early pregnancy loss, and live birth in both fresh and frozen cycles. Serum hCG-β levels of POD 12 and 14 and the fold increase of hCG-β levels from POD 12 to 14 significantly predict pregnancy outcomes after fresh and frozen cycles. Among these, the cutoff value of POD 14 hCG-β had the highest sensitivity and positive predictive value (PPV). In fresh cycles, the cutoff values of POD 12 and 14 serum hCG-β levels for clinical pregnancies were 30.2 mIU/ mL (sensitivity 81.3 %, specificity 79.6 %, and PPV 92.3 %) and 70.5 mIU/mL (sensitivity 88.4 %, specificity 85.2 %, and PPV 94.7 %). In pregnancies with POD 12 serum hCG-β levels ≥30.2 mIU/mL, the cutoff level of increase of hCG-β for clinical pregnancy was 2.56 (sensitivity 73.6 %, specificity 72.4 %, and PPV 97.8 %). Sequential application of cutoff values such as POD 12 hCG-β and fold increase of hCG-β improved predictability of pregnancy outcome as compared with that of POD 12 hCG-β alone. The cutoff values of POD 12 and 14 serum hCG-β levels for live birth were 40.5 mIU/ mL (sensitivity 75.2 %, specificity 72.6 %, PPV 78.9 %) and 104.5 mIU/mL (sensitivity 80.3 %, specificity 74.1 %, PPV 80.8 %). In the frozen cycles, the cutoff values of POD 12 and 14 serum hCG-β level for clinical pregnancy were 31.5 IU/L (sensitivity 80.4 %, specificity 71.1 % and PPV 90 %) and 43.5 mIU/mL (sensitivity 72.6 %, specificity 71.7 %, PPV 77.2 %). In pregnancies with POD 12 serum hCG-β level ≥31.5 mIU/mL, the cutoff value for fold increase of hCG-β was 2.38 for clinical pregnancy (sensitivity 81.6 %, specificity 71.4 % and PPV 87.9 %). The cutoff values of POD 12 and 14 for live birth were 43.5 mIU/mL (sensitivity 72.6 %, specificity 71.7 %, PPV 77.2 %) and 101.6 mIU/mL (sensitivity 79.6 %, specificity 71.1 %, PPV 78.4 %). Sequential application of cutoff values for POD 12 hCG-β level and fold increase of hCG-β significantly increased PPV for live birth but not clinical pregnancy in frozen cycles. Conclusions Early prediction of pregnancy outcome by using POD 12 and 14 cutoff levels and sequential application of cutoff value of fold increase could provide appropriate reference to health care providers to initiate earlier management of Capsule Early prediction of pregnancy outcome...