N-(2-chloroethyl)-N-ethyl-2-bromobenzylamine (DSP-4) is a potent and highly selective neurotoxin which induces degeneration of noradrenergic axons. The effects of DSP-4 vary considerably in different brain regions: the drug produces nearly complete depletion of noradrenaline in neocortex, hippocampus, cerebellum and spinal cord, but only partial depletion in hypothalamus and brainstem. In this study we have employed an immunohistochemical method to assess the neurotoxic effects of DSP-4 on the structural integrity of central noradrenergic neurons in the rat, and to identify those noradrenergic axons that remain in the central nervous system 2-4 weeks after DSP-4 treatment. The staining results identified noradrenergic axon terminals as the principal site of action of DSP-4; noradrenergic cell bodies and preterminal axons were not noticeably affected. DSP-4 produced an almost all or none neurotoxic effect on noradrenergic axon terminals in different brain regions. Nearly all noradrenergic axon terminals were destroyed in the neocortex, hippocampus, olfactory bulb, thalamus, tectum, cerebellum and spinal cord dorsal horn. In contrast, most noradrenergic axons were unaffected in the basal forebrain, hypothalamus, reticular formation, brainstem motor nuclei and spinal cord ventral horn. These remaining noradrenergic axon terminals differed morphologically from sensitive axons by their thickness, size and spacing of their varicosities and their dense arborizations within terminal fields. The distribution of noradrenergic axons susceptible to DSP-4 correlates very closely with the distribution of locus coeruleus axons and possibly all regions in which noradrenergic terminals are unaffected by DSP-4 receive their major noradrenergic input from non-locus coeruleus neurons. This study provides the first direct evidence that DSP-4 destroys noradrenergic axon terminals from the locus coeruleus, but not those from non-locus coeruleus neurons. This profound differential sensitivity of noradrenergic axons to DSP-4 is matched by distinct differences in their morphology and their topographic projections. The results support the view that locus coeruleus and non-locus coeruleus noradrenergic neurons constitute two separate subsystems, which differ not only in their projections but also with respect to the pharmacological properties of their axon terminals.