Branched and isoprenoid glycerol dialkyl glycerol tetraethers (GDGTs) are membrane lipids of bacteria and archaea, respectively, and their core lipid distributions are used as proxy indicators in paleolimnological studies. In addition, the amount and composition of intact polar lipid (IPL) GDGTs yield information on the presence and abundance of GDGT‐producing microbes within the water column. GDGTs are, however, not always easily recovered from cultured microbial cells by commonly applied extraction methods, and this may also apply to suspended particulate matter (SPM) in aquatic systems. In order to investigate potential biases induced by incomplete GDGT recovery, we analyzed both core‐ and IPL‐GDGTs in SPM from Lake Lugano (Switzerland) using (1) ultrasonic solvent extraction with mixtures of methanol (MeOH) and dichloromethane (DCM), and (2) two modifications of the Bligh–Dyer (BD) protocol. Acid hydrolysis of the post‐extraction residues revealed that particularly branched GDGTs were poorly recovered from SPM by the MeOH/DCM mixtures (25–62%). Much better extraction yields (> 85%) were achieved with the BD method, however, during subsequent phase separation, up to 75% of the extracted branched IPL‐GDGTs partitioned into the aqueous phosphate buffer that is usually discarded. In contrast, when this buffer was substituted with 5% trichloroacetic acid, only 13% were lost into the aqueous phase. Depending on the protocol used, the distribution of the IPL‐derived GDGTs varied substantially. Our results indicate that both bacterial‐ and archaeal IPL‐GDGTs can be difficult to extract from lacustrine microbial communities, and caution is advised when targeting the “viable” GDGT pool.