Direct homing refers to the behaviour whereby an ant with food runs in an almost straight line to the nest. We determined whether non-polarized light acts as a directional cue in the direct homing behaviour of the grass ant, Tetramorium caespitum. We carried out tests (1) under a blue sky or in total overcast conditions in the field, and (2) with a view of the sun or with a non-polarized light within a box in the laboratory. The ants' paths were recorded, and then analysed by means of circular and linear statistics. Ant workers with food were able to home directly under a blue sky, under totally overcast skies, and even under non-polarized lighting. In a fifth test we displaced the ants at the beginning of the homing trip; they returned to the presumed nest location along a path that ran parallel to the true homing vector, as if they had not been displaced. The results suggest that the ants home by dead reckoning, that is, they can measure directional changes and distances travelled during the outward trip, and integrate them into a mean home vector for direct homing depending on an external light-based reference system, that is, either the polarized skylight pattern or non-polarized lighting. Copyright 1998 The Association for the Study of Animal Behaviour. Copyright 1998 The Association for the Study of Animal Behaviour.