Since the early attempts of Bowen to account for the diversity of the igneous rocks, a number of proposals have been offered to explain the voluminous andesites of the continental orogenic belts. These include: (1) contamination of basic magmas;(2) anatexis of sialic crust; (3) differentiation of basaltic magmas; (4) anatexis of wet peridotite mantle; and (5) anatexis of oceanic crust and mantle in subduction zones. This paper assembles some of the results of endeavors in experimental petrology and geochemistry to evaluate these proposals. The concept of contamination is supported by neither geologic occurrence nor isotope geochemistry. Similar judgements, together with the unrealistically high temperatures that are required, argue against anatexis of sialic crust to produce primary andesitic liquids. Differentiation of basic magmas to produce andesitic and other silica-rich melts unquestionably occurs. Recent experimental evidence on the hydrothermal melting of basalts reveals that fractionation of silica-poor amphiboles is probably a more efficacious mechanism for producing the calc-alkaline trend than is the early separation of magnetite. Nevertheless, it has yet to be demonstrated that such processes have operated to the extent necessary to produce the orogenic andesites. High-pressure melting experiments on wet synthetic peridotites disclose that quartz-normative liquids are the products of partial melting at pressures up to about 20 kbar. This suggests that andesites can develop from anatexis of water-bearing continental and oceanic mantle to depths of about 60 km, but this mechanism alone fails to account for the spatial distribution of the orogenic calc-alkaline suite. Armed with experimental data on melting and other phase relationships of basic and ultrabasic compositions, together with recent data on the abundances of trace and minor elements in andesites, the most attractive models involve anatexis of mantle peridotite or oceanic crust metamorphosed to amphibolite or eclogite in subducted lithosphere. Melting may also occur in rocks above the hanging wall of the Benioff zone as the result of incorporation of water released by dehydration of the descending slab. The fractionation of alkalis into amphiboles, and into micas at greater depths, may be responsible for the observed zoning of K 2O in surficial calc-alkaline rocks overlying the distal plate boundaries. However, this model is not in accord with the concept of generation of calc-alkaline magmas at depths of hundreds of kilometers proposed by some workers.