Behavioural bioassays have shown that worker honeybees can distinguish between worker-laid and queen-laid eggs. By eating worker-laid eggs, nest-mate workers pre-vent each other from reproducing, a behaviour known as worker policing. However, the recognition cue used by worker honeybees to discriminate between worker-laid and queen-laid eggs remains elusive. This study describes observations made and experiments conducted to elucidate the nature of the recognition cue. No visually apparent physical differences were found between worker-laid and queen-laid eggs using scanning electron micrographs (SEMs) magnified up to 2500 times, which indi-cates that the cue is probably chemical in nature. We confirmed that the signal prob-ably resides on the queen-laid and not the worker-laid eggs, since a worker-laid egg is protected when placed in contact with a queen-laid egg. We conducted a series of standard egg-removal bioassays in queenright colonies using queen-laid and worker-laid eggs treated with a wide range of potential recognition chemicals, solvents, buff-ers or gland extracts. The aim was either to disrupt or remove the egg-marking signal from queen-laid eggs, or to add the signal to worker-laid eggs. Despite a comprehen-sive set of experiments, we were unable to alter the egg-marking signal on queen-laid eggs or transfer the signal from queen-laid eggs to worker-laid eggs. Furthermore, two candidate signals, esters from queens and eicosenol from workers, were shown not to be the cues used by workers. This indicates that the egg-marking signal in honeybees is remarkably robust and consists of a chemical or group of chemicals not previously associated with chemical signalling in social insects.