Some workers have held that mid-ocean ridge basalts are fractionated from high pressure (15–30 kbar) picritic primary magmas whereas others have favored primary magmas generated at about 10 kbar with compositions close to those of mid-ocean ridge basalts. Of critical significance are presumed differences in composition between experimentally determined primary magmas and the least fractionated mid-ocean ridge basalts. To evaluate the significance of these differences, all based on electron microprobe analyses, we consider three sources of uncertainty: (1) analytical uncertainties for a single microprobe laboratory, (2) systematic interlaboratory analytical differences, and (3) real variations in the possible compositions of primary magmas that can be produced from a peridotite source at a given pressure. The first source of error is surprisingly large and can account for a substantial part of the total variation of normative quartz (hypersthene calculated as equivalent olivine and quartz) in FAMOUS basalts. The second is not as serious but remains undetermined for many laboratories. The third is potentially the largest but is not yet fully documented. The least fractionated FA-MOUS basalts have high mg numbers (70–73) compatible with derivation from the mantle by direct partial melting with little or no subsequent fractional crystallization. Because of the wide range of normative quartz content in these basalts, it appears necessary to consider them as representatives of multiple parental magmas. When all the sources of uncertainty are taken into account, we conclude that the experimental data by various investigators are all fairly consistent and favor derivation of the least fractionated mid-ocean ridge basalts by at most only a small amount of fractional crystallization from primary magmas having a wide range of normative quartz content and generated over a range of pressures from about 7–11 kbar.