We present a detailed ab initio investigation of the stability, the structural, electronic, and magnetic properties of the (0001) surfaces of hematite $({\mathrm{Fe}}_{2}{\mathrm{O}}_{3})$ and chromia or eskolaite $({\mathrm{Cr}}_{2}{\mathrm{O}}_{3})$. Strong electron correlation effects not included in a density-functional description are described by a Hubbard-type on-site Coulomb repulsion (the $\mathrm{DFT}+U$ approach). For bulk chromia we find, complementing our recent work on hematite [Rollmann et al., Phys. Rev. B 69, 165107 (2004)] that the inclusion of correlation effects leads to an improved description of the structural, electronic, and magnetic properties. In particular, the increased exchange splitting of the $d$ band changes the character of the insulating gap from a pure $d\text{\ensuremath{-}}d$ Mott-Hubbard type to intermediate between $d\text{\ensuremath{-}}d$ and charge-transfer insulator. For both oxides, the strong correlation effects have a dramatic influence on the surface stability: oxygen-terminated surfaces are strongly disfavored because of the increased energetic cost of stabilizing a higher oxidation state of the transition metal close to the surface. The stability of metal-terminated surfaces even under oxidizing conditions agrees with the most recent STM and LEED data. For ${\mathrm{Cr}}_{2}{\mathrm{O}}_{3}(0001)$ where detailed experimental information on the surface structure is available, quantitative agreement of the calculated surface relaxations is achieved. Detailed results on the surface electronic structure (valence-band spectra and core-level shifts) and the surface magnetic properties are presented.