A standardized data workflow is described for large-scale serum metabolomic studies using multisegment injection-capillary electrophoresis-mass spectrometry. Multiplexed separations increase throughput (<4 min/sample) for quantitative determination of 66 polar/ionic metabolites in serum filtrates consistently detected (coefficient of variance (CV) <30%) with high frequency (>75%) from a multi-ethnic cohort of pregnant women (n = 1,004). We outline a validated protocol implemented in four batches over a 7-month period that includes details on preventive maintenance, sample workup, data preprocessing and metabolite authentication. We achieve stringent quality control (QC) and robust batch correction of long-term signal drift with good mutual agreement for a wide range of metabolites, including serum glucose as compared to a clinical chemistry analyzer (mean bias = 11%, n = 668). Control charts for a recovery standard (mean CV = 12%, n = 2,412) and serum metabolites in QC samples (median CV = 13%, n = 202) demonstrate acceptable intermediate precision with a median intraclass coefficient of 0.87. We also report reference intervals for 53 serum metabolites from a diverse population of women in their second trimester of pregnancy.