Four major phases are distinguished during the building of the Pacaya volcanological complex (Guatemala): (1) the ‘ancestral volcano’, now much eroded, covered by younger deposits and battered by faulting and landslides; (2) the initial cone made up of large lava flows and dated at about 0.5 Ma; (3) andesito-dacitic domes (Cerro Chiquito dome and others) emplaced during an extrusive phase at about 0.16 Ma; and (4) the active Pacaya volcano. Lavas of phases 2 and 4 are basalts and basaltic andesites with almost the same major and trace element compositions. Classical enrichment in LILE and depletion in HFSE are observed. Phase 3 domes show magma-mingling features. The dacitic host rock includes basaltic andestic enclaves, 20 to 30% in volume. According to geochemical and mineralogic data (Mg/Fe ratios of basic minerals higher in dacite, groundmass glasses sodic in dacite and potassic in basaltic andesite), the basaltic andesites and dacites of phase 3 cannot be related by a simple fractional crystallization process. The existence of such differences suggests that magma mingling/mixing processes were involved by a connection between the two magma chambers prior to the extrusion of the andesito-dacitic domes. However, some trace element data clearly suggest that fractional crystallization played a significant role in the differentiation of these lavas. Remelting of amphibole-bearing cumulates from the dacite may also have played a role in the basaltic andesitic liquid genesis. Thermodynamical parameters of each liquid are contrasted. The basaltic andesitic magma, at a high temperature (1037°C) and in relatively small amounts, is embayed in the cooler (905° C) dacitic magma. The former liquid, denser (2.72) and less viscous (103.31 poises for free crystal liquid) may crystallize while the latter, lighter (2.60) and more viscous (104.46 poises), remains still liquid. Isotopic data (0.70383<87Sr/86Sr <0.70400; 0.512785<143Nd/144Nd<0.512908; 18.61<206Pb/204Pb<18.66; 15.56<207Pb/204Pb <15.58; 38.30<208Pb/204Pb<38.40) indicate that all the lavas (from Pacaya as well as from Cerro Chiquito) are cogenetic and derive from the same mantle source. Sr, Nd and Pb isotope ratios are similar to those of OIBs. (230Th/232Th) activity ratios on two historical lavas are respectively 1.2 and 1.3. The Th excess is similar to that of other calcalkaline volcanoes emplaced on a continental crust. These lavas evolved, possibly in separate magma chambers, through processes of fractional crystallization and magma mixing.