Experimentally determined pyroxene phase relations at 800¿-1200 ¿C are combined with calculated phase equilibria for the Di-En and Hd-Fs joins to produce a graphical two-pyroxene thermometer that should be applicable to a wide variety of rocks from the earth, moon, and meteorites. The thermometer can be used directly with natural pyroxenes that have low contents of Al and other minor components. Samples with higher contents of 'others' components require special projection onto the Ca-Mg-Fe pyroxene quadrilateral; Wo, En, and Fs as normally calculated will not yield correct temperatures! The special projection is necessary to approximate the activities of those components in natural pyroxenes. The effects of pressure are nonnegligible but can be corrected for. Use of the thermometer for slowly-cooled rocks may pose special problems if the pyroxenes have undergone granule exsolution (coalescence of exsolved material to form separte grains). For example, published analyses of Mg-rich augites from the Skaergaard intrusion give temperatures that are 50¿-100 ¿C below expected magmatic values. Textural and experimental evidence confirms that these augites have undergone granule exsolution. The primary pyroxene compositions must be reconstructed from textural evidence before correct igneous or peak-metamorphic temperatures can be obtained from this or any two-pyroxene thermometer. Of the numerous two-pyroxene thermometers proposed in the literature, those of Ross and Huebner <1975> and of Kretz <1982; Ca-reaction> yield temperatures most similar to our own, although the former tends to give temperatures about 50 ¿C lower for some igneous pyroxenes. Other thermometers generally overestimate temperatures for metamorphic rocks (hornblende granulite facies) by 50¿-100 ¿C.