We provide an extensive set of theoretical spectral energy distributions of massive stars derived from our "combined stellar structure and atmosphere models". The calculations cover the entire main sequence evolution for initial masses M_i = 20 - 120 M_sun, corresponding to O3-B0 stars of all luminosity classes.

We predict detailed line blanketed UV spectra along the main sequence evolution. The major result is a systematic study of ionizing fluxes covering the entire parameter space of O and early B stars. We demonstrate the importance of accounting simultaneously for non-LTE effects, line blanketing and stellar winds to obtain an accurate description of the spectra of these stars shortward of the Lyman limit. The main results from our spectra are the following:

The flux in the HeII continuum is increased by 2 to 3 (3 to 6) orders of magnitudes compared to predictions from plane parallel non-LTE (LTE) model atmospheres. This reconfirms the work of Gabler et al. (1989).
The flux in the HeI continuum is known to be increased due non-LTE effects. However, we find that it is also influenced by wind effects as was previously found by Najarro et al. (1996) and Schaerer et al. (1996b). The combined effect of a mass outflow and line blanketing leads to a flatter energy distribution in the HeI continuum, which confirms the results of Sellmaier et al. (1996) for a wider range of stellar parameters.
The flux in the Lyman continuum is also modified due to line blanketing and the presence of a stellar winds, although to a lesser degree than the spectrum at higher energies.

We derive revised ionizing fluxes for O3 to B0 stars based on the recent temperature and gravity calibrations of Vacca et al. (1996). The total number of Lyman continuum photons is found to be slightly lower than previous derivations. For most cases the differences are less than ~ 20 %. Due to the increased flux in the HeI continuum the hardness ratio of the HeI to H continuum is increased by ~ 1.6 to ~ 2.5 depending on spectral type and luminosity class.

In the view of recent EUV and X-ray observations, a critical discussion of current model assumptions (including our own) shows that for stars of spectral types later than approximately B0, which have relatively weak stellar winds, reliable predictions of ionizing fluxes are not yet possible. We identify the most likely physical reasons for this finding.