Tyrosine hydroxylase-immunoreactive fibres in the rat neostriatum were studied in the electron microscope in order to determine the nature of the contacts they make with other neural elements. The larger varicose parts of such fibres contained relatively few vesicles and rarely displayed synaptic membrane specializations; however, thinner parts of axons (0.1-0.4 micron) contained many vesicles and had symmetrical membrane specializations, indicative of en passant type synapses. By far the most common postsynaptic targets of tyrosine hydroxylase-immunoreactive boutons were dendritic spines and shafts, although neuronal cell bodies and axon initial segments also received such input. Six striatonigral neurons in the ventral striatum were identified by retrograde labelling with horseradish peroxidase and their dendritic processes were revealed by Golgi impregnation using the section-Golgi procedure. The same sections were also developed to reveal tyrosine hydroxylase immunoreactivity and so we were able to study immunoreactive boutons in contact with the Golgi-impregnated striatonigral neurons. Each of the 280 immunoreactive boutons examined in the electron microscope displayed symmetrical synaptic membrane specializations: 59% of the boutons were in synaptic contact with the dendritic spines, 35% with the dendritic shafts and 6% with the cell bodies of striatonigral neurons. The dendritic spines of striatonigral neurons that received input from immunoreactive boutons invariably also received input, usually more distally, from unstained boutons that formed asymmetrical synaptic specializations. A study of 87 spines along the dendrites of an identified striatonigral neuron showed that the most common type of synaptic input was from an individual unstained bouton making asymmetrical synaptic contact (53%), while 39% of the spines received one asymmetrical synapse and one symmetrical immunoreactive synapse. It is proposed that the spatial distribution of presumed dopaminergic terminals in synaptic contact with different parts of striatonigral neurons has important functional implications. Those synapses on the cell body and proximal dendritic shafts might mediate a relatively non-selective inhibition. In contrast, the major dopaminergic input that occurs on the necks of dendritic spines is likely to be highly selective since it could prevent the excitatory input to the same spines from reaching the dendritic shaft. One of the main functions of dopamine released from nigrostriatal fibres might thus be to alter the pattern of firing of striatal output neurons by regulating their input.