Serum estradiol (E2) concentrations in midreproductive women are easily measured using a variety of conventional immunoassays (IA). However, when women approach and traverse menopause, E2 eventually drops below levels where IA lacks sufficient sensitivity to accurately measure E2. Liquid chromatography and tandem mass spectrometry (LC/MS/MS) has become the standard method for assessing steroid hormones, especially when circulating concentrations are low. We evaluated the relationship between IA and LC/MS/MS E2 measurements in a cohort of women taken from the Study of Womens Health Across the Nation (SWAN) to assess the degree of agreement between the two methods and to determine the level of E2 at which IA becomes unreliable. Methods: 315 serum samples that had been previously measured for E2 using IA were re-analyzed using LC/MS/MS performed by one of the authors (RA). In this original set, E2 levels that were below the limit of assay detection (LLD, 6 pg/ml) were interpolated as a random number between 0 and the LLD. Agreement between all 315 samples was assessed using both Pearson and Spearman correlation. The analysis was repeated excluding the subset of specimens that were below the lower limit of detection (LLD) for the IA E2 assay (6 pg/ml; N=176), and a third set of correlations was obtained for specimens that measured Results: The overall dataset (N=315) demonstrated excellent agreement between IA and LC/MS/MS with a Pearson’s r and Spearmans r of 0.98 AND 0.60, respectively. When the subset of 176 samples above the LLD were assessed, Pearson’s r was 0.98 and Spearman’s r was 0.81. In contrast, when specimens measuring 6–15 pg/ml by IA were compared to LC/MS/MS, Pearson’s r was -0.03 and Spearman’s r was 0.09, indicating a complete loss of relationship between the two methods. Conclusions: The IA used by SWAN (England, Clin Chem 2002; 48: 1584) and LC/MS/MS demonstrate excellent correlation for E2 measurements above 15 pg/ml. However, circulating concentrations of E2 below 15 pg/ml were not accurately measured using IA.