For the quantitative determination of paclitaxel in human plasma reversed-phase high-performance liquid chromatographic (HPLC) methods with solid-phase extraction (SPE) as sample pretreatment procedure are frequently used. Recovery problems arose during the quantification of paclitaxel in plasma samples of patients. The major problems were a large batch-to-batch difference in performance of the SPE columns and the effects of the pharmaceutical vehicle Cremophor EL on the performance of the SPE. Cremophor EL concentrations exceeding 1.0% (v/v) had a great impact on the absolute recovery of paclitaxel from human plasma with the SPE procedure. The recoveries decreased approximately 10 to 40% depending on the quality of the batch SPE columns. The problems are avoided by using 2'-methylpaclitaxel as the internal standard. This study points out the importance of including the effects of a pharmaceutical vehicle, like Cremophor EL, in the validation programme of a bioanalytical assay and the use of an internal standard in HPLC paclitaxel assays preceded by SPE as sample pretreatment procedure.