Breathing is governed by the coordinated activities of brainstem neurons located in three main clusters: the pontine, dorsal and ventral respiratory groups, and at other sites, e.g. the retrotrapezoid nucleus, the medullary raphe and the locus ceruleus. This control system initiates the breathing rhythm and maintains a pattern of breath size and frequency that achieves appropriate gas exchange at minimal energetic cost. To do so it integrates afferent input from multiple mechano- and chemoreceptors and coordinates motor output to different muscles for inspiration, expiration and airway resistance. Our understanding of this system is far from complete (Feldman et al. 2003). We do not know how the breathing rhythm begins, nor do we know the full chemical phenotype of breathing-related neurons at different sites, or how breathing-related neurons among these sites interact to maintain the breathing pattern. We do not know how neuronal function might be modified by ongoing activity (plasticity). The absence of non-lethal effects on breathing in chronic, conscious animals following substantial lesions in many parts of this system hints at redundancy and complexity. Evolving experimental strategies will help to understand this apparently simple yet enigmatic control system. Further knowledge of neuronal chemical phenotype will allow dissection of the role of specifically identified neurons. For example, cell specific killing by injection of a conjugate of the toxin saporin with substance P kills only neurokinin-1 receptor-expressing neurons with dramatic effects on breathing at a putative site important in rhythm generation (Gray et al. 2001) and at a central chemoreceptor site (Nattie & Li, 2002). Single gene mutations allow insight into the role of neurons that express that gene, which may be widely dispersed. For example, mice deficient in monoamine oxidase A express excess serotonin and unstable breathing in vitro (Bou-Flores et al. 2000). Mice deficient in a developmental transcription factor Pet-1 have few serotonergic neurons (Hendricks et al. 2003) and could provide further insight into the role of these neurons in breathing. Carefully conceived studies using traditional approaches remain useful. In principle all neurons express receptors for excitatory amino-acids, which allows for focal chemical stimulation in order to examine function and map location. As reported in this issue of The Journal of Physiology, Monnier et al. (2003) used microinjections of dl-homocysteic acid (DLH), an excitatory amino acid, to examine the micro-circuitry within the ventral respiratory group of breathing-related neurons. This work is noteworthy in its execution. They applied very small injections of low doses (the effective size of each injection is within a ≈300 μm radius) at precisely determined locations to describe focal sites with breathing responses in the rostral to caudal extent of this neuron group. The study is remarkable for the thorough and systematic nature of their approach, the careful attention to small injection size, the measure of multiple nerve outputs and the detailed mapping at appropriate inter-injection distances. Their extensive results allow them to put more fragmented, and sometimes contradictory, data of others into a reasonably consistent framework and to argue persuasively for a functional compartmentalization of the ventral respiratory group neurons, an important advance. They find four separate regions each approximately 500 μm in length, which when stimulated produce different effects on fictive breathing rate, rhythm and amplitude and on blood pressure. Each of three regions provides unique input for breathing rate or rhythm; all three have a similar effect on fictive breath size. The study provides information at high resolution of an anatomical and functional nature necessary in order to understand system organization and function. Future application of this approach together with cell specific manipulation via genetic or chemical means will expand its utility and our knowledge of this control system.