Untargeted food lipidomics is increasingly applied in food research. Yet the impact of pretreatment and lipid extraction methods on the lipidomics profiles detected in food matrices has not been systematically characterized. We used the brown macroalgae Fucus vesiculosus as a model to examine how these factors shape the lipid recovery and lipidomic profiles detected using ultra-high performance liquid chromatography-high resolution quadrupole time of flight mass spectrometry (RPLC–QTOF–MS/MS). Within Bligh & Dyer (BD) extraction system, homogenization time, ultrasonication time and extraction cycles significantly altered total lipid yield and the number of detected lipids. Homogenization and ultrasonication as sample pretreatments improved recovery of glycolipids and phospholipids, whereas prolonged treatments led to the loss of highly unsaturated polar lipids. Increasing the water phase in the solvent system had little effect on lipid yield but reduced the number of polar lipid species detected. Under identical pretreatments, BD and Folch extractions yielded similar total numbers of lipid species (∼270) but clearly different class distributions: Folch extraction favoured TG and SQDG, whereas BD extraction enriched MGDG and several betaine-type lipids. These results show that both pretreatment conditions and choice of extraction methods can markedly influence lipid yields, polar lipid coverage, and relative abundance of different lipid classes. Careful design and transparent reporting of extraction protocols are therefore essential for robust and comparable food lipidomics.