The galaxy IRAS F10214 + 4724, discovered in a spectroscopic survey of a 0.2-Jy sample by Rowan-Robinson and collaborators in 1991, is significantly more luminous than any other known galaxy. Its bolometric luminosity is 2x10(14) L., which is comparable to the luminosities of the most luminous quasars. Recent observations have revealed a candidate foreground group of galaxies, which might gravitationally lense F10214 + 4724, thus explaining much of its luminosity. High-resolution imaging of F10214 + 4724 has revealed that most of its near-IR flux comes from a circularly symmetric are; this also supports the gravitational lens interpretation. In such a scenario, F10214 + 4724 would be the high-redshift analogue of the ultraluminous IRAS galaxies observed locally. This work presents a simple statistical lensing model to investigate this possibility. We show that, on statistical grounds alone, the probability that F10214 + 4724 is a gravitational lens system with magnification 2 < mu < 10 is approximately 25 per cent, If nearby determinations of the luminosity function phi(L) for ultraluminous IRAS galaxies can be extrapolated to both high redshifts z and high luminosities L. If phi(L) steepens with either increasing z or increasing L, we predict a substantial increase in the probability of F10214 + 4724 being a lens system. Very large magnifications (mu > 20) are ruled out by this model, unless phi(L) is very steep, e.g., a power law with index a < - 6. These results therefore suggest that F10214 + 4724 is indeed the most luminous galaxy known. However, if it is a lens system with mu > 2, it would not have been discovered had it not been lensed.