Testing the safety and efficacy of a successful human medicine involves many laboratory animals, which can sometimes be subjected to considerable suffering and distress. Also, it is necessary to extrapolate from the test species to humans. UK and European legislation requires that Replacement, Reduction and Refinement of animal procedures (the Three Rs) are implemented wherever possible. Over the last decade, there has been substantial progress with applying in vitro and in silico methods to both drug efficacy and safety testing. This paper is a report of the discussions and recommendations arising from a workshop on the role that might be played by human volunteer studies in the very early stages of drug development. The workshop was organised in November, 2001 by Volunteers in Research and Testing, a group of individuals in the UK which launched an initiative in 1994 to identify where and how human volunteers can participate safely in biomedical studies to replace laboratory animals. It was considered that conducting pre-Phase I very low dose human studies (sub-toxic and below the dose threshold for measurable pharmacological or clinical activity) could enable drug candidates to be assessed earlier for in vivo human pharmacokinetics and metabolism. Moreover, accelerator mass spectrometry (AMS), nuclear magnetic resonance (NMR) spectroscopy and positron emission tomography (PET) are potentially useful spectrometric and imaging methods that can be used in conjunction with such human studies. Some, limited animal tests would still be required before pre-Phase I microdose studies, to take account of the potential risk posed by completely novel chemicals. The workshop recommended that very early volunteer studies using microdoses should be introduced into the drug development process in a way that does not compromise volunteer safety or the scientific quality of the resulting safety data. This should improve the selection of drug candidates and also reduce the likelihood of later candidate failure, by providing in vivo human ADME data, especially for pharmacokinetics and metabolism, at an earlier stage in drug development than is currently the case.