OBJECTIVE Measurements of serum progesterone to predict early gestational normalcy have been found to be as predictive as serial hCG titers. Since ultrasound would be the diagnostic tool of choice if hCG was > 2000 mIU/ml, the purpose of the present study was to determine the best predictive value of a single progesterone measurement when hCG levels were < 2000 mIU/ml. METHODS Relative operating characteristic analysis of progesterone level as a predictor of early gestational normalcy when hCG is < 2000 mIU/ml. METHODS Ninety-three pregnant patients that conceived spontaneously were evaluated with progesterone measurements when the patient's hCG was < 2000 mIU/ml. Two-by-two contingency tables were constructed that compared pregnancy outcome with multiple discriminatory serum progesterone concentrations between 0 and 38 ng/mL. From these tables, a relative operating characteristic (ROC) curve was generated to compare the sensitivity and false-positive rates. RESULTS Of a total of 93 pregnancies, 27 had a normal outcome and 66 had an abnormal outcome. The ROC curve indicated that a serum progesterone concentration of 12 ng/ml had the highest sensitivity associated with the lowest false-positive rate. The area under the curve was equal to 0.941 +/- 0.024. This observation was compared to our previously reported data of progesterone levels that included hCG levels > 2000 mIU/ml, yielding an area under the curve of 0.772 +/- 0.053. Calculation of the critical ratio z revealed that there is a significant improvement in the predictive value of progesterone when hCG is < 2000 mIU/ml (P < 0.005). CONCLUSIONS A single serum progesterone level has a better predictive value of pregnancy normalcy when hCG measurements are < 2000 mIU/ml.