Reserpine + nialamid administration to the rat induces a strong yellow fluorescence of the neuronal bodies of the raphe, due to serotonin (5-HT) accumulation. Under these conditions, administration of clomipramine (an antidepressant drug acting preferentially on 5-HT-mediated neurons) induces a decrease of intraneuronal fluorescence and its interneuronal diffusion. On this pattern we administered new antidepressant drugs which act on 5-HT neurons in a much more intensive way than clomipramine (fluvoxamine, clovoxamine, LM 5008, citalopram, Ro 11-2465). To varying degrees, we observed in the raphe, in addition to a decrease in intraneuronal fluorescence and interneuronal diffusion, the presence of a yellow fluorescence in capillary walls. It seems that under these antidepressants, 5-HT, which is outside neuronal bodies because of uptake blockade, is partly caught by the capillary walls. In these walls rich in monoamine oxydase, 5-HT would be catabolized, 5HIAA dispersed in the blood and thus, this 'capillary effect' could correspond to a loss of 5-HT in the raphe. Antidepressant drugs preferentially acting upon noradrenaline (NA) neurons do not, in this model, induce analogous phenomena in NA cell bodies of the locus coeruleus. So the 'capillary effect' differentiates antidepressant drugs acting specifically on 5-HT or NA neurons. It may be considered together with other parameters which also indicate asymmetries on the modes of action of antidepressant drugs, such as effects on monoamine turnover (increase for NA and decrease for 5-HT) and on receptor sensitivity (decrease for NA and increase for 5-HT).