This paper reports the first use of non-LTE, line-blanketed model atmospheres to analyze the spectra of hot stars. The stars analyzed are members of clusters and associations, have spectral types in the range O9-B2 and luminosity classes in the range III-V, have slow to moderate rotation, and are photometrically constant. The model-atmosphere code, PAM, has been described by Anderson in 1985 and 1987. In the present study, sampled line opacities of iron-group elements were incorporated into the radiative transfer solution. Solar abundances were assumed. Synthetic line profiles of H-alpha, H-beta, H-gamma, and He I lambda-lambda-4471 and 4922 were fitted to the stellar spectra in order to choose T(eff), log g, and v sin i for each star. In most of the program stars, we obtained good to excellent agreement between the computed profiles and essentially all the line profiles used to fix the model, and we derived reliable stellar parameters. With other lines, however, the models demonstrate less success in reproducing observed line strengths, probably due to inadequate model ions for He II, N II, and C II. The synthetic Mg II lambda-4481 equivalent widths agree well with the observed ones at the low end of the temperature range studied, but, above 25,000 K, the synthetic line is generally stronger than the observed line. In H-alpha, the fit to the inner line wings is improved over what was achieved in previous studies, but the present study finds that the synthetic line cores are systematically too deep. The cause of this discrepancy is unknown. For most stars in common with previous studies, there are no disagreements in the derived stellar parameters that can be attributed to the physics included in the model atmosphere. In 10 Lacertae, a disagreement with the effective temperature of Schonberner et al. can be attributed to the choice of lines to which the model was fitted. The sample of stars includes delta-Scorpii, for which the angular diameter has been measured. In common with other model-atmosphere investigations, we find an effective temperature slightly lower than that implied by the angular diameter. We did not use the models to compute color indices, but we constructed the relation between the observed Stromgren c0 index and our derived values of T(eff). The result is a useful temperature calibration for c0. The behavior of the observed equivalent widths of N ii, N III, C II, and C III lines as a function of T(eff) was studied. Most of the lines show much scatter, with no consistent trend that could indicate abundance differences from star to star. Even the claimed N star, HD 12323, does not show a consistent enhancement in the strengths of the N lines, although the data do support a C deficiency in this star. The predicted behavior of He I lambda-5876 does not conform to its observed behavior, which can therefore be presumed to be dominated either by abundance differences from star to star or by some physical process that is not included in the model atmosphere. The predicted behavior of He II lambda-4686 is qualitatively as observed, but the observations show a great deal of scatter at a given T(eff). Since the line is strongly temperature sensitive, however, most of the scatter may be due to temperature errors.
