In molecular beam epitaxy an optical pyrometer is frequently used to determine substrate and epilayer temperatures during the growth of semiconductors. This implies a calibration of the pyrometer and a correct analysis of its indications. In this article we discuss problems presented by this calibration. We first compare the temperatures measured on the sample holder and on several substrates (GaAs, InAs, GaSb) attached to it. Then we present examples of temperature variations recorded during the growth of AlSb and GaSb on GaSb or GaAs substrates. GaSb epilayers thicker than 0.1 μm induce a 40°–60° decrease of the measured temperature. AlSb deposition on GaSb produces temperature oscillations. We discuss these variations in terms of apparent and true temperature variations.