New electron and ion microprobe analyses of 36 olivines, 9 pyroxenes, and 10 garnets included in diamond were obtained. Olivine inclusions from kimberlites with Precambrian ages have higher mean Cr2O3 (850 versus 360 ppmw), Al2O3 (200 versus 90 ppmw), CaO (490 versus 210 ppmw), Na2O (110 versus 30 ppmw), and Li (686 versus 466 counts per second) than olivine inclusions found in Mesozoic kimberlites, perhaps indicating a temporal change of either temperature or chemistry or both in the mantle source regions of diamonds. In contrast to Iherzolitic olivines, aluminium and sodium in olivines from diamonds are insufficient to couple with trivalent chromium; perhaps divalent chromium is inferred. Some olivines show zoning to Ca-rich edges. Temperature and pressure estimates for inclusions in four diamonds fall in the combined pressure-temperature range of coarse and porphyroclastic garnet lherzolites, and temperature estimates for inclusions in seven diamonds lacking a pressure estimate cover most of the temperature range fo lherbolites. Taken at face value, the pressure and temperature estimates for inclusions in diamonds indicate that some diamonds might have crystallized in the presence of a liquid containing H2O, CO2, or both. Other estimates fall in the subsolidus field even for H2O-saturated peridotite. Although the pressure-temperature estimates overlap for lherzolites and ultramafic-type inclusions in diamonds, differences in trace element content of olivines require that the crystallization environments were not identical. Perhaps the Cr content of the olivine inclusions indicates that the host diamonds grew in a reducing environment not found in typical lherzolites.