Alzheimer's disease (AD) is an irreversible neurodegenerative disorder with the clinical symptom of the progressive loss of cognitive function and mild cognitive impairment (MCI) is a translational state between cognitive changes of normal aging and AD. Lipid metabolism and pathogenesis of Alzheimer's disease (AD) are closely linked. Despite obviously discrete lipidome constitutions across lipoproteins, lipidomic approaches of AD has been mostly conducted without considering lipoprotein-dependent alterations. This study introduces a combination of asymmetrical flow field-flow fractionation (AF4) and nanoflow ultrahigh performance liquid chromatography-tandem mass spectrometry (nUHPLC-ESI-MS/MS) for a comprehensive lipid profiling in different lipoprotein level of patients plasma with AD and amnestic MCI in comparison to age-matched healthy controls. Lipoproteins in plasma samples were size-sorted by a semi-preparative scale AF4, followed by non-targeted lipid identification and high speed targeted quantitation with nUHPLC-ESI-MS/MS. It shows 14 significantly altered high abundance lipids in AD, exhibiting >2-fold increases (p < 0.01) in LDL/VLDL including triacylglycerol, ceramide, phosphatidylethanolamine, and diacylglycerol. Three lipid species (triacylglycerol 50:1, diacylglycerol 18:1_18:1, and phosphatidylethanolamine 36:2) showing a strong correlation with the degree of brain atrophy were found as candidate species which can be utilized to differentiate the early stage of MCI when simple Mini-Mental State Examination results were statistically incorporated. The present study elucidated lipoprotein-dependent alterations of lipids in progression of MCI and further to AD which can be utilized for the future development of lipid biomarkers to enhance the predictability of disease progress.