Fluid inclusion studies give unique insights into physical conditions and composition of fluids involved in geological processes. Most studies to date are performed on transparent minerals. Near-infrared (NIR) microscopy allows also microthermometry to be performed on minerals that are opaque to the visible light such as pyrite, hematite, wolframite, enargite and stibnite. The main drawback of this technique is the underestimation of the recorded phase-change temperatures with increasing light intensity, up to several hundred percent in the case of ice-melting temperatures. Although this issue has been known for a decade, it is poorly understood. We address this problem based on a systematic study of synthetic fluid inclusions in a variety of opaque minerals. For the first time, fluid inclusions have been co-synthetized with success in quartz and opaque minerals. Fracturing the host minerals by in-situ quenching allowed for fluid–mineral equilibration prior to fluid inclusion formation. In this study, we assess the impact of mineral intrinsic parameters, mainly absorption and thermal-conductivity, and experimental settings (light source operative power, diaphragms aperture, and the use of filters) on recorded phase-change temperatures. We show that these are underestimated due to local overheating of the sample caused by radiative heating from the light source. It affects all minerals and the extent of the temperature shift of the observed phase-changes depends on sample thickness, mineral characteristics, and the amount of light reaching the sample. Thus, any calibration of the temperature shift as a function of the amount of light is complicated and in most cases impracticable. However, we demonstrate, based on co-generated inclusions in opaque minerals and quartz that yield similar values during NIR-microthermometry that for any mineral, there is a range of light power and microscope settings for which no shift is noticeable within the thermal stage accuracy. These are defined as "ideal measuring conditions" ensuring reliability of acquired microthermometry data.