It is obvious that there are great problems with pharmacokinetic individualization of anticancer therapy. The strong relationship between dose intensity (total dose/unit time) and response revealed in clinical trials with some tumours provides a strong support for studies seeking relationships between the individual plasma pharmacokinetic profile and response to treatment. Unfortunately, studies that define a therapeutic window are sparse, and trials that prospectively test such models are even rarer. Thus, for most cancer drugs, it is not possible to give any definite advice on how to use pharmacokinetic determinations to establish individualised therapy, and there is therefore a definite need for such studies. It is important, however, that attempts to establish relationships between drug concentrations and therapeutic effects be founded on a sound theoretical base. When drugs, mainly antimetabolites, are extensively metabolised intracellularly and interact with intracellular processes about which there are data showing a strong interindividual heterogeneity, such data must be considered when designing pharmacokinetic investigations. Cytarabine and fluorouracil are good examples of this. The monitoring of intracellular drug/metabolite concentrations or of the direct biochemical events in the tumour cells seems to be a promising approach with such drugs. It also needs to be emphasised that pharmacokinetically guided individualization cannot be achieved before a therapeutic window is established, i.e. a knowledge of the relationship between drug concentration and clinical effects. The investigators in this field accept a great responsibility when clinical studies are undertaken: a poorly designed study showing no benefit from pharmacokinetically guided individualization can impair the possibilities of performing more adequate studies in the future.