A timing analysis of VHE gamma-ray observations of Cen X-3 taken with the University of Durham Mark 6 telescope was performed, using both Rayleigh and Bayesian statistics. Data from only 1 observation produced evidence for periodicity, in a 'soft' cut data set, with an overall chance probability <0.01 and blueshifted by 6.6ms from the nominal second harmonic of the pulse period. We show that the disappearance of the Rayleigh peak in the fully cut data set is consistent with a gamma-ray origin hypothesis. The second part of the paper considers theoretical models that could explain the existing data of Cen X-3 in high and very high energy gamma-rays. On the basis of energetics, all reasonable options for gamma-ray production must be connected to jets emerging from the inner accretion disc around the neutron star. One principal option is a large scale source, with R_s~10^13-14cm, assuming effective acceleration of electrons by shocks produced by the interaction of the jets with the dense atmosphere of the binary. The bulk of gamma-ray features could be explained by this model except modulations at the pulsar period, which would require a source with R_s<10^11cm. We consider hadronic and leptonic models for the formation of such a compact object. Both models predict the episodes of pulsed emission may be rare, of typical duration ~few hours, and the frequency of pulsations should be significantly shifted from the nominal frequency of the X-ray pulsar. The opportunities to distinguish between different models using future gamma-ray observations of this X-ray binary are discussed. (v. abridged)