Using (1)H-nuclear magnetic resonance (NMR) spectroscopy and statistical models, we sought to identify ''biomarkers'' present in erythrocytes that would distinguish between women with normal pregnancy and those suffering from preeclampsia, and investigate possible links with previously identified plasma ''markers.'' Erythrocytes from 22 normotensive pregnant women and 15 preeclamptics were analyzed by (1)H Carr-Purcell-Meiboom-Gill (CPMG) NMR. Multivariate analysis and logistic regression were applied to differentiate between the 2 groups of patients, and used to develop a diagnostic model based on the concentrations of the constituents identified as being influential. Significantly higher concentrations of alanine (P < .001), glycine (P = .025), and ergothioneine (P = .049) were found in erythrocytes from preeclamptic patients. Discriminant analysis and regression of NMR data permitted 100% accurate diagnosis of the health status of new patients. Chemically related imidazole-based molecules, histidine and ergothioneine, are important in the classification process and the etiology of preeclampsia (PE).