<p>Organic aerosols (OA) are complex mixtures comprising thousands of compounds, posing challenges for molecular characterization. Liquid chromatography coupled with high-resolution mass spectrometry (LC-HRMS) is a powerful tool for the analysis of OA. The selection of extraction solvents critically influences the molecular characterization of OA in LC-HRMS, yet selection biases are poorly understood. In this study, we systematically evaluated the extraction efficiencies of three commonly used solvents—methanol (MeOH), acetonitrile (ACN), and an ACN/H₂O mixture (8:2, v/v)—for diverse OA samples using HPLC-Orbitrap MS/MS. The sample set encompassed seasonal ambient PM_2.5 collected in a Chinese megacity, and aerosols emitted from cooking, coal combustion, and biomass burning. Generally, MeOH was the optimal solvent for non-targeted screening of ambient aerosols, yielding the broadest compound coverage and the highest signal intensities. However, ACN exhibited superior extraction efficiency for cooking and coal combustion aerosols, showing distinct selectivity towards compounds with higher carbon number (nC), higher DBE and lower OSc, particularly for aromatic and nitrophenolic compounds. ACN/H₂O extracted compounds with lower nC, DBE and higher OSc. It showed higher signal intensity for species containing hydrophilic functional groups, such as amines, pyridines, carbonyl compounds, polycarboxylic acids, and carbohydrates. This selectivity indicates that solvent selection may introduce relative quantification biases for specific compounds. The evaluation of procedural contamination revealed that ACN is most severely affected by procedural contaminants, particularly for long-chain fatty acid amides, aromatic and aliphatic amines from polypropylene plasticware. This study provided critical insights for optimizing solvent selection in non-targeted LC-HRMS analysis.</p>