In the spontaneous mutant Ames dwarf mouse, GH deficiency coincides with a dramatic increase in the expression of both mRNA and peptide for stimulatory GHRH and reduced expression of GH-inhibitory somatostatin (SRIH) mRNA and peptide. However, both GHRH and SRIH are markedly reduced in the dwarf median eminence (ME), suggesting that ME innervation by GHRH and SRIH neurons may be aberrant in the absence of GH. In order to test this hypothesis, the number of GHRH and SRIH ME-projecting neurons was evaluated in normal and dwarf mice using a combination of retrograde tract-tracing and neuron phenotype identification by immunocytochemistry (ICC). Adult animals were injected intraperitoneally with the fluorescent tract-tracer fluorogold (FG), which, in the brain, is taken up only by axons terminating in areas deprived of the blood-brain barrier, such as the ME. Visualization of FG was achieved by either UV illumination or ICC, and was combined as appropriate with fluorescence or bright-field ICC for GHRH or SRIH. Cells immunoreactive for GHRH or SRIH and labeled with FG were quantified at each 180-µm rostral-to-caudal level through the hypothalamus. As reported previously, the total number of hypophysiotropic GHRH neurons was markedly increased in dwarf compared with that in normal mice. However, a similar percentage of ME-innervating GHRH cells was estimated in dwarf (73 ± 4%) and normal (76 ± 3%) animals. Such a percentage in dwarfs thus represents a larger population of ME-projecting GHRH cells (749 ± 53) than in normal mice (128 ± 15). Increased numbers of FG-labeled GHRH neurons in dwarfs were located at the middle and posterior levels of the arcuate nucleus (2.08, 2.26 and 2.44 mm posterior to bregma). The percentage of FG-labeled SRIH neurons was also similar for dwarf (83 ± 2%) and normal (87 ± 2%) mice. Because the total SRIH-immunoreactive neuronal population in dwarfs is significantly reduced compared to that in normal animals, the similar FG-labeled percentage reflects a reduced number of SRIH cells projecting to ME in dwarf (1,376 ± 104) compared with normal (3,192 ± 267) mice. Fewer FG-labeled SRIH cells were found in dwarfs at every anterior-to-posterior level of the periventricular nucleus (p < 0.01 for comparisons at 0.28, 0.46, 0.64, and 1.0, and p < 0.05 for comparison at 1.18 mm posterior to the bregma). The present study indicates that the reduction in GHRH and SRIH immunoreactivity in the dwarf ME may result from different phenomena for each neuronal population. The reduction in GHRH immunostaining in the ME, despite a marked increase in the total ME-projecting GHRH neurons, may be interpreted as increased GHRH release, with consequent depletion of the ME stores. In contrast, the deficit in ME SRIH may be proportional to the deficit in the number of detectable SRIH periventricular nucleus neurons.