Currently, there is an increasing focus on the implementation of pharmacokinetic-pharmacodynamic (PK-PD) studies and modelling as essential tools for drug development. Strategies involving specifically the population approach, which are based on relatively recent statistical methodology (e.g. nonlinear mixed effects modelling, NONMEM) have been advocated for investigating pharmacokinetic and pharmacodynamic variability as well as dose-concentration-effect relationships. The present article outlines this approach, and discusses how it can be implemented within the framework of the studies currently performed as part of the clinical phases of new drug development. It also considers study design and performance, based on real-life experiences. Population approaches, if designed carefully and early, as part of the planning of the drug development programme, are expected to play a significant role at every phase of the programme and to contribute to providing information that is valuable for registration purposes. Statistical methodology and software are now widely available. However, practical issues such as integration of the population approach within existing protocols, quality control of the data, timing of laboratory and statistical analyses, as well as resource allocation, remain legitimate concerns to be considered in prospective studies.