Blood sample preparation before LC-MS metabolomic fingerprinting is one of the most challenging and error-prone parts of the analytical procedure. Besides proteins, phospholipids contained in blood fluids are known to cause matrix effects and ion suppression phenomena, thus masking biological variation. Nevertheless, the commonly used sample preparation techniques do not consider their removal prior to analysis. Pooled plasma and serum samples were used as biological material, partly as raw samples and partly spiked with distinct concentrations of a metabolite mix (1-5 μg/mL). Prior to LC-ESI-qToF-MS-driven metabolomic analysis, samples were subjected to different preparation methods consisting of three extractions with organic solvents (acetonitrile, methanol, and methanol/ethanol), a membrane-based solvent-free technique, and a hybrid method combining solvent extraction and SPE-mediated removal of phospholipids. The comparative analysis among sample preparation procedures was based on the capacity to detect endogenous compounds in raw samples, differentiate raw versus spiked samples, and reveal real-life metabolomic changes, following a dietary intervention. Method speed, minimum sample handling, compatibility to automation, and applicability to large-scale metabolomic studies were also considered. The combination of solvent deproteinization and the selective removal of phospholipids was revealed to be the most suitable method, in terms of improvement of nonlipid metabolite coverage, extraction reproducibility, quickness, and compatibility with automation, the minimization of matrix effects being among the most probable causes for the good extraction performance associated with the removal of phospholipid species. The main advantage of conventional solvent extraction procedures was the metabolite information coverage for lipid low-molecular-weight species, and extraction with acetonitrile was generally the second choice for sample preparation. Ultrafiltration was the least effective method for plasma and serum preparation; thus, its use without a previous solvent extraction step of the samples should be discarded. According to the presented data, there is no apparent reason to believe that sacrificing information on lipid compounds is too high of a price to pay in order to gain more information on nonlipid LMW metabolites.