How do honeybees use visual odometry and goal-defining landmarks to guide food search? In one experiment, bees were trained to forage in an optic-flow-rich tunnel with a landmark positioned directly above the feeder. Subsequent food-search tests indicated that bees searched much more accurately when both odometric and landmark cues were available than when only odometry was available. When the two cue sources were set in conflict, by shifting the position of the landmark in the tunnel during test, bees overwhelmingly used landmark cues rather than odometry. In another experiment, odometric cues were removed by training and testing in axially striped tunnels. The data show that bees did not weight landmarks as highly as when odometric cues were available, tending to search in the vicinity of the landmark for shorter periods. A third experiment, in which bees were trained with odometry but without a landmark, showed that a novel landmark placed anywhere in the tunnel during testing prevented bees from searching beyond the landmark location. Two further experiments, involving training bees to relatively longer distances with a goal-defining landmark, produced similar results to the initial experiment. One caveat was that, with the removal of the familiar landmark, bees tended to overshoot the training location, relative to the case where bees were trained without a landmark. Taken together, the results suggest that bees assign appropriate significance to odometric and landmark cues in a more flexible and dynamic way than previously envisaged.