We derive the bolometric to X-ray correlation for a local sample of normal and starburst galaxies and use it, in combination with several UV reddening schemes, to predict the 2--8 keV X-ray luminosity for a sample of 24 Lyman-break galaxies in the HDF/CDF-N. We find that the mean X-ray luminosity, as predicted from the Meurer UV reddening relation for starburst galaxies, agrees extremely well with the Brandt stacking analysis. This provides additional evidence that Lyman-break galaxies can be considered as scaled-up local starbursts and that the locally derived starburst UV reddening relation may be a reasonable tool for estimating the UV extinction at high redshift. Our analysis shows that the Lyman-break sample can not have far-IR to far-UV flux ratios similar to nearby ULIGs, as this would predict a mean X-ray luminosity 100 times larger than observed, as well as far-IR luminosities large enough to be detected in the sub-mm. We calculate the UV reddening expected from the Calzetti effective starburst attenuation curve and the radiative transfer models of Witt & Gordon for low metallicity dust in a shell geometry with homogeneous or clumpy dust distributions and find that all are consistent with the observed X-ray emission. Finally, we show that the mean X-ray luminosity of the sample would be under predicted by a factor of 6 if the the far-UV is unattenuated by dust.