We compared two HPLC methods (anion exchange [AE] and steric exclusion [SE]) for analysis of mouse lipoprotein profiles by determining coefficients of variability (CVs) under varying conditions. CVs for AE and SE were comparable on fresh samples. There was an inverse relationship between subfraction curve area and CV [r = -0.65 (AE) and -0.50 (SE)], consistent with the interpretation that as curve area decreased, error variance increased and signal-to-noise ratio decreased. Sample storage did not affect SE. In contrast, with AE, alterations in measured lipoproteins were apparent after storage, including a decrease in the HDL subfraction [66.8% (baseline) vs. 15.9% (1 week); P < 0.01] and an increase in areas under LDL and VLDL peaks. Concomitant with decreasing HDL area, reproducibility deteriorated with the duration of storage. Analysis of the effects of decreasing sample injectate volume to <25 microl on SE lipoprotein subfractions revealed that areas under LDL and VLDL peaks decreased and persisted as volume was decreased further. Areas under all lipoprotein subfractions measured with either AE or SE were linearly correlated with the amount of cholesterol [r = 0.69 (AE) and 0.87 (SE)]. Both AE and SE yield reproducible, accurate, and rapid measurements of lipoproteins from small amounts of serum. AE yields more sensitive, high-amplitude, well-defined peaks that can be easily distinguished and necessitates the use of smaller sample volumes compared with SE, but sample storage causes alterations in the chromatogram. SE appears better suited to serial analyses involving stored samples.