The use of complementary and alternative medicines (CAM) by cancer patients is increasing. Concomitant use of CAM and anticancer drugs could lead to serious safety issues in patients. CAM have the potential to cause pharmacokinetic interactions with anticancer drugs, leading to either increased or decreased plasma levels of anticancer drugs. This could result in unexpected toxicities or a reduced efficacy. Significant pharmacokinetic interactions have already been shown between St. John's Wort (SJW) and the anticancer drugs imatinib and irinotecan. Most pharmacokinetic CAM-drug interactions, involve drug metabolizing cytochrome P450 (CYP) enzymes, in particular CYP3A4. The effect of CAM on CYP3A4 activity and expression can be assessed in vitro. However, no data have been reported yet regarding the relevance of these in vitro data for the prediction of CAM-anticancer drug interactions in clinical practice. To address this issue, a literature research was performed to evaluate the relevance of in vitro data to predict clinical effects of CAM frequently used by cancer patients: SJW, milk thistle, garlic and Panax ginseng (P. ginseng). Furthermore, in clinical studies the sensitive CYP3A4 substrate probe midazolam is often used to determine pharmacokinetic interactions. Results of these clinical studies with midazolam are used to predict pharmacokinetic interactions with other drugs metabolized by CYP3A4. Therefore, this review also explored whether clinical trials with midazolam are useful to predict clinical pharmacokinetic CAM-anticancer drug interactions. In vitro data of SJW have shown CYP3A4 inhibition after short-term exposure and induction after long-term exposure. In clinical studies using midazolam or anticancer drugs (irinotecan and imatinib) as known CYP3A4 substrates in combination with SJW, decreased plasma levels of these drugs were observed, which was expected as a consequence of CYP3A4 induction. For garlic, no effect on CYP3A4 has been shown in vitro and also in clinical studies garlic did not affect the pharmacokinetics of both midazolam and docetaxel. Milk thistle and P. ginseng predominantly showed CYP3A4 inhibition in vitro. However, in clinical studies these CAM did not cause significant pharmacokinetic interactions with midazolam, irinotecan, docetaxel and imatinib. Most likely, factors as poor pharmaceutical availability, solubility and bioavailability contribute to the lack of significant clinical interactions. In conclusion, in vitro data are useful as a first indication for potential pharmacokinetic drug interactions with CAM. However, the discrepancies between in vitro and clinical results for milk thistle and P. ginseng show that clinical studies are required for confirmation of potential interactions. At last, midazolam as a model substrate for CYP3A4, has convincingly shown to correctly predict clinical interactions between CAM and anticancer drugs.