Ferrario & Wickramasinghe (2006) explored the hypothesis that the magnetic fields of neutron stars are of fossil origin. In this context, they predicted the field distribution of the progenitor OB stars, finding that 5 per cent of main sequence massive stars should have fields in excess of 1kG. We have carried out sensitive ESPaDOnS spectropolarimetric observations to search for direct evidence of such fields in all massive B- and O-type stars in the Orion Nebula Cluster star-forming region. We have detected unambiguous Stokes V Zeeman signatures in spectra of three out of the eight stars observed (38%). Using a new state-of-the-art Bayesian analysis, we infer the presence of strong (kG), organised magnetic fields in their photospheres. For the remaining five stars, we constrain any dipolar fields in the photosphere to be weaker than about 200G. Statistically, the chance of finding three ~kG fields in a sample of eight OB stars is quite low (less than 1%) if the predictions of Ferrario & Wickramasinghe are correct. This implies that either the magnetic fields of neutron stars are not of fossil origin, that the flux-evolution model of Ferrario & Wickramasinghe is incomplete, or that the ONC has unusual magnetic properties. We are undertaking a study of other young star clusters, in order to better explore these possibilities.