Using the flocking tendencies of homing pigeons (Columba livia) in flight, a first attempt was made to test directly the hypothesis that group movement may have a navigational function. We manipulated the potential advantage of transfer of navigational information by releasing pairs of homing pigeons in which none, one or both of the birds had previously been released at sites about 25 km from the home loft (while overall flying experience was standardized). Contrary to our prediction, locally naive birds did not follow experienced birds home, whereas experienced birds, with less navigational need to flock, were generally still paired on arrival home. Pairs in which both birds were locally naive also tended not to home together. Under the conditions of our experiment, the tendency to fly together appeared not to have a navigational function. However, and contrary to recent conclusions in the literature, our results do suggest that the initial flight behaviour of birds at novel release sites may constitute a navigational information gathering strategy that is sufficiently important to over-ride gregarious tendencies when the two conflict.