1. The activity recorded from mammalian sympathetic nerves comes in bursts, which result from large numbers of fibres firing synchronously. 2. Human sympathetic nerve activity behaves similarly to that in animals, although burst rates may be lower. 3. Vasomotor, cardiac and sudomotor nerve fibres all fire in bursts. Whether other sympathetic pathways do so is unknown. 4. Sympathetic activity is intrinsically 'bursty' but not intrinsically regular. 5. Bursting is a population phenomenon, not usually evident in the firing of individual neurons. 6. Bursts in post-ganglionic nerves are driven by synchronously firing preganglionic neurons. 7. The origin of bursts remains controversial. Preganglionic neuron properties are likely to be important in at least shaping bursts. 8. Burst amplitude, which reflects the number of fibres firing together, and burst probability are controlled independently. 9. Baroreceptors affect burst probability over a wide range, but have less effect on mean burst amplitude. How they affect burst timing within the cardiac cycle is discussed. 10. Burst probability is determined 'downstream' of the rostral ventrolateral medulla, implicating either the spinal cord or recurrent brainstem connections in burst generation. 11. Neuroeffector responses are too slow to follow individual bursts. However, bursting will promote spatial facilitation at both ganglionic and effector levels, which may increase the dynamic range of neural control.