Although microdialysis is commonly understood as a method of sampling low molecular weight compounds in the extracellular compartment of tissues, this definition appears insufficient to specifically describe brain microdialysis of neurotransmitters. In fact, transmitter overflow from the brain into dialysates is critically dependent upon the composition of the perfusing Ringer. Therefore, the dialysing Ringer not only recovers the transmitter from the extracellular brain fluid but is a main determinant of its in-vivo release. Two types of brain microdialysis are distinguished: quantitative micro-dialysis and conventional microdialysis. Quantitative microdialysis provides an estimate of neurotransmitter concentrations in the extracellular fluid in contact with the probe. However, this information might poorly reflect the kinetics of neurotransmitter release in vivo. Conventional microdialysis involves perfusion at a constant rate with a transmitter-free Ringer, resulting in the formation of a steep neurotransmitter concentration gradient extending from the Ringer into the extracellular fluid. This artificial gradient might be critical for the ability of conventional microdialysis to detect and resolve phasic changes in neurotransmitter release taking place in the implanted area. On the basis of these characteristics, conventional microdialysis of neurotransmitters can be conceptualized as a model of the in-vivo release of neurotransmitters in the brain. As such, the criteria of face-validity, construct-validity and predictive-validity should be applied to select the most appropriate experimental conditions for estimating neurotransmitter release in specific brain areas in relation to behaviour.