Background Toxicity and therapeutic effects of paclitaxel vary greatly between patients and remain a clinically relevant problem with regard to the handling of dose delay/reduction or termination of treatment. We investigated the notion that single nucleotide polymorphisms (SNPs) in CYP2C8 could be partly responsible for this variation. Paclitaxel is mainly metabolized by CYP2C8; SNPs have been investigated in this context before but conclusions are still lacking. We present a prospective study of paclitaxel clearance (CL) in 93 Caucasian females with epithelial ovarian cancer with regard to the CYP2C8 *1b, *1c, *3 and *4 genotypes. Material and methods All patients were diagnosed with primary ovarian/peritoneal cancer and received 175mg/m2 paclitaxel over 3 hrs plus carboplatin (AUC5-6) q3w. All patients gave written and verbal consent. The study was approved by ethics committees in Denmark and Sweden. Blood was sampled at 3hrs, 5-8 hrs and 18-24hrs after start of infusion. Total plasma paclitaxel was quantified using HPLC. CremophorEL® (CrEL) was determined as described by Sparreboom et al.1998. CL of unbound paclitaxel was estimated using total concentrations, CrEL and other parameters in the model described by Henningsson et al.2003 using NONMEM VI. Genotypes were determined using Pyrosequencing. Genotypes were in HW equilibrium, except *1b (p=0.01). Results The PK model predicted the data well. The CL of unbound paclitaxel was lower for patients with the CYP2C8*3 and *4 variants (p<0.05). CYP2C8 variant N Paclitaxel CL geometric mean (l/h) 95% c.i P-value (T-test of log transformed CL) Wt/Wt 49 395.0 [370.3;421.4] Wt/*1b + *1b/*1b(n=1) 44 374.5 [348.8;402.2] 0.267 Wt/Wt 69 382.4 [362.1;404.0] Wt/*1c 24 393.1 [355.4;434.8] 0.617 Wt/Wt 74 394.7 [375.2;415.3] Wt/*3 19 350.0 [310.0;395.3] 0.041* Wt/Wt 86 390.9 [372.1;410.7] Wt/*4 7 320.8 [281.7;365.3] 0.028* Note: for the one patient carrying both the *3 and *4 variant CL was 269.7 l/h Conclusions This study implies reduced elimination of paclitaxel in Caucasian female patients with the CYP2C8*3 and *4 genotypes. This confirms several in vitro studies and pilot studies but is different to Henningsson et al 2005 and Marsh et al 2006 which could be explained by differences in dose ranges, infusion times and/or related to gender. The finding is important in terms of understanding inter individual variability of paclitaxel pharmacokinetics and might in the future provide useful information for individualized chemotherapy.