A re-evaluation of kinetic data shows that the methoxide base-catalyzed methanolysis of soybean oil at 40 degrees C (6:1 methanol:oil molar ratio) to form methyl esters proceeds approximately 15 times more slowly than butanolysis at 30 degrees C. This is interpreted to be the result of a two-phase reaction in which methanolysis occurs only in the methanol phase. Low oil concentration in methanol causes the slow reaction rate; a slow dissolving rate of the oil in the methanol causes an initiation period. Intermediate mono- and diglycerides preferentially remain in the methanol, and react further, thus explaining the deviation from second-order kinetics. The same explanations apply for hydroxide ion catalyzed methanolysis. At the 6:1 methanol:oil molar ratio the addition of a cosolvent, such as 1.25 volumes of tetrahydrofuran (THF) per volume of methanol, produces an oil-dominant one-phase system in which methanolysis speeds up dramatically and occurs as fast as butanolysis. The critical separation of the glycerol-rich phase still occurs and does so faster than in the cosolvent-free system. For THF, recycle of solvent is simplified because of the similar boiling points of THF (67 degrees C) and methanol (65 degrees C). Possible explanations for the abnormal slowing of the methanolysis reactions are presented in terms of (1) lower rate constants for mono- and diglyceride reactions due to the formation of cyclic intermediates, (2) a fall in the polarity of the reaction mixture due to either methanol depletion or mixing of the oil, methanol and cosolvent, and (3) depletion of hydroxide ion when this is present. Copyright (C) 1996 Elsevier Science Ltd.