The crystallization of chromium-bearing spinel from silicate magmas is first examined, and then an attempt is made to apply the theory developed in Irvine (1965) to the principal occurrences of the mineral for which data are available. It is concluded that the chromium-rich varieties commonly known as chromite have generally formed simultaneously with olivine, and that their crystallization has in many occurrences been terminated by a peritectic (reaction) relation leading to formation of a pyroxene. The origin of the aluminium-rich varieties is more problematical and perhaps more varied; however, they evidently occur only in alpine-type peridotite bodies and peridotite "nodules" in basaltic volcanic rocks, and their formation may generally have involved high pressures. It is found that there is a rough correlation between the Mg/Fe++ ratios of chromian spinels, olivines, and pyroxenes occurring in the same rock bodies, and there is some evidence that the Mg–Fe++ distribution coefficients of spinel–silicate pairs may significantly be sensitive to temperature. The chromites of stratiform intrusions reflect an appreciable range of oxygen fugacities, whereas the spinels of alpine- type peridotite bodies seem generally to have formed at about the same relatively low oxygen fugacity. Chromites from ultra mafic bodies of the type common to southeastern Alaska have exceptionally high Fe+++/Cr + Al + Fe+++ ratios, probably because the bodies crystallized from magma that was extremely poor in silica.