A number of different approaches to the study of functional neurochemistry in human brain are discussed. The advantages and disadvantages of three main techniques are contrasted: (i) using animal tissue preparations as models of the human brain; (ii) using human peripheral tissue preparations as models of dynamic CNS processes; and (iii) studying human tissue, obtained postmortem, directly. Animal models are often readily obtained and reliable, and the high degree of inbreeding of common laboratory animals ensures that they usually yield consistent results. However, there are a number of human disorders for which animal models are either poor or unavailable, and species differences make extrapolation from the animal to the human case difficult. Human peripheral tissue models rely on a degree of homology between peripheral and CNS processes; in most cases, the evidence for such homologies derives from animal, rather than human, studies. Moreover, several examples are known where a peripheral process mimics the equivalent glial cell activity more closely than the neuronal, which can be a serious drawback for studies of neurotransmission. The use of postmortem human brain tissue presents a number of obvious difficulties, resulting from variations in the patient's age, agonal state, sex, preterminal medication, postmortem delay, etc. Human beings are genetically and nutritionally heterogeneous, so that data variability is usually greater here than when using tissue from laboratory animals. However, it is possible to control for a number of these factors, for example, by matching samples for basal metabolic rate and tissue integrity, and recently developed tissue freezing and storage techniques permit the use of within-subject experimental designs to help reduce experimental variation. A range of neurotransmitter functions are well retained in such tissue samples, so that regional variations, differential transmitter activities, drug effects, etc., can be studied in normal tissue samples, as well as in samples taken from cases of neurological and psychiatric disease. This allows, for example, changes in neuroanatomical indices to be correlated with localised alterations in a specific neurotransmitter function. A systematic approach to the analysis and matching of tissue samples is advocated. The three approaches should be considered to be complementary, especially for the study of human brain diseases.