Systematic experiments on the cave-fishes Anoptichthys jordani and Caecobarbus geertsi show that the vertical orientation of swimming behaviour can be significantly modified by feeding the fishes during a period of 8 days before the experiments, either with food dropping to the bottom (Tubifex) or with food floating on the surface of the water (dried Daphnia). The observations have been quantified by recording systematically the position of the fishes in height in an aquarium divided in 3 levels of observation. The position of the fishes was recorded every 10 seconds during a total period of 15 minutes, divided in 3 periods of 5 minutes each (Minutes 0 to 5: anterior to stimulation (AS); Minutes 5 to 10: first period after stimulation (PS1); Minutes 10-15: second period after stimulation (PS2)). Each type of observation has been repeated 4 times at intervals of several days. The median values of the 5 minutes observation periods have been compared by means of the Mann-Whitney U test (alpha = 0,05). The orientation changes resulting from systematic feeding with floating food, show that the stereotype of polarisation on the substrate, which exists in Anoptichthys and, with some differences, in Caecobarbus, is a rather variable behavioural character. This movement is more pronounced in Anoptichthys than in Caecobarbus. This difference can be theoretically interpreted in relation with the fact that Caecobarbus is a more ancient cave-form than Anoptichthys. According to this hypothesis, the stereotype of polarisation on the substrate would be more liable to be modified in Anoptichthys, due to a higher variability of behaviour structures, the latter being more rigid in an ancient form in relation with phyletic degeneration processes typical of true troglobiotic animals. Similarly, systematic feeding with floating material does not cause any preferential reaction for the upper water level before chemical stimulation in Caecobarbus, the reverse being true for Anoptichthys. Mechanical stimulation induces phobic reactions whose dominant vertical orientation is determined, in Anoptichthys, by the type of feeding material used during the pre-experimental period. This does not hold for Caecobarbus, who shows a definite preference for the substrate in these conditions. It is remarkable that, in spite of the more rigid polarisation on the substrate shown by Caecobarbus, the preferential orientation to the lower level is not followed by an active exploratory behaviour as in Anoptichthys. This dissociation between substrate polarisation and exploratory behaviour is to be interpreted, once more, as a sign of deeper phyletic degeneration in the ethology of Caecobarbus. In Anoptichthys the effects of the group seem to favour the preferential reaction for the vertical level at which food is present, whereas in Caecobarbus, the presence of specific mates is rather inhibitory.