New determinations of effective temperatures of 23 magnetic, chemically peculiar (mCP) stars were obtained from a fit of metal‐enhanced model atmospheres to the observed spectral energy distributions (SED) from ultraviolet (UV) to red. Temperatures of four more CP stars without magnetic measurements were also obtained, of which three show light variations characteristic of magnetic variables, and the fourth may possibly be normal. The root‐mean‐square (rms) method was used to fit the theoretical SED to the observations corrected for reddening if necessary, with metallicity and effective temperature as the fitting parameters. Gravity was assumed to be equal to log g= 4 for main‐sequence stars and to log g= 3 for two giants in the considered sample. Equal weights were given to the UV part and visual part of SED. An attempt was made to obtain a three‐parameter fit (with reddening included) to the observed SED of HD 215441. It turned out that the three‐parameter fit gave spurious results due to correlations among the parameters. We could determine the reddening of that star only by assuming that the metallicity of the model is known, hence reducing the number of fitting parameters to two. Independently of the formal quality of fit resulting from the rms method applied to the whole SED, the quality of fit was additionally checked for each star by determination of the temperature from the best‐fitting model atmosphere to the UV part and the visual part of SED separately. Both temperatures should be close to one another if the global best‐fitting model satisfactorily describes the full observed SED. This is the case for about a half of the investigated stars but the difference exceeds 750 K for the remaining stars with the extreme values above 2000 K. Possible reasons for such discrepancies are discussed. New revised calibrations of effective temperature and bolometric corrections of mCP stars in terms of reddening free Strömgren indices are given.