Four species of nocturnal trans-saharan passerine migrants (Ficedula hypoleuca, Phylloscopus trochilus, Phoenicurus phoenicurus, Sylvia borin) were tested in Emlen funnels to assess the relative importance of the setting sun and stars in determining the nocturnal course. The experiments took place during spring migration at a Mediterranean stop-over site (Sardinia, Italy). The birds were caught during the day in mist-nets and kept in opaque plastic cages preventing them seeing outside, until the time of the experiment, on the same night; each bird was tested only once. Two experimental groups were exposed to the sunset, the first including the sight of the sun disc, the other after the sun had set. A third group was kept in the closed cages until the time of the test. A fourth group was directly tested in the funnels at sunset time, after the sun disc had set. The first three groups were tested in the funnels later, in complete darkness. Exposure and tests occurred under the same sky conditions, either clear or overcast. Under a clear sky, 10 out of 16 distributions are oriented in the expected migratory direction, or have a tendency towards it, irrespective of their exposure to the setting sun. This shows that the cues present at sunset are not necessary for the orientation of these birds, whose ability to orient correctly in the absence of any celestial cues is also confirmed by the results of the tests performed in overcast conditions with no previous exposure to visual cues. Unexpectedly, exposure to the setting sun with an overcast sky, followed by a test in the same overcast conditions lead to disorientation, suggesting that exposure to visual cues makes the birds look, at least at first, for other similar cues before turning to non-visual compass information.