The sensitivity of complex analytical procedures depends not only on the sensitivity of the analytical instrument used, but also on the recovery degree of the examined analyte by the employed sample preparation method. The recovery degrees of individual cannabinoids reported in literature, estimated using the same sample preparation method, are unexpectedly divergent. Therefore, the aim of this study was a thorough assessment of the most commonly used sample preparation methods, such as protein precipitation, LLE, QuEChERS and SPE, in the context of the reliability of the obtained results. The presented report shows that the highest sensitivity, precision and reliability of the chromatographic analysis of CBG, CBD, ∆9-THC and CBN in human plasma can be obtained using SPE. The recovery degrees of these cannabinoids by SPE are highly repeatable and exceed 95 %, while they are significantly lower for such sample preparation methods as protein precipitation, LLE and QuEChERS (ca. 80, 65 and 87, respectively). Moreover, the supernatants obtained by the latter methods contain interferents evoking matrix-effect, which makes reliable quantification of the listed cannabinoids by GC difficult. To our knowledge, the paper is the first such extensive comparison of sample preparation procedures used for the determination of cannabinoids in plasma by GC-MS and HPLC-MS. The presented results and the discussion allow to understand why different recovery degrees for the same xenobiotic can be find in literature despite they have been estimated using the same or different sample preparation method or different chromatography types. Display omitted •Analyte recovery degree as a key factor in analytical procedure of complex mixtures.•SPE gives almost 100 % recovery of cannabinoids from blood plasma.•SPE allows for reliable analysis of cannabinoids in blood/plasma by GC and HPLC.•Plasma LLE and QuEChERS supernatants contain interferents evoking matrix effect in GC.