The Bureau of Mines has carried out experimental and theoretical studies with prilled and pulverized ammonium nitrate-fuel oil (an- fo) mixtures containing varying amounts of fuel oil in an attempt to quantify the effects of stoichiometric composition, nonideal detonation behavior, and expansion volume on the production of CO, no, and NO2 fumes. Experimental fume measurements were obtained in the Bureau's large closed gallery facility (7.2 X 104 liter expansion chamber) and in the standard Crawshaw-Jones Apparatus (90- liter expansion chamber) using a prepackaged charge configuration containing about 450 g of explosive. The theoretical calculation of toxic fumes was achieved with an equilibrium detonation code called "tiger." Contrary to initial expectations, the nox (= no + NO2) fumes from the large gallery test were found to be in essential agreement with the Crawshaw-Jones results. It was also concluded that (1) "tiger" calculations offer a good approach to the prediction of toxic fumes; (2) there is a basic problem in extrapolating laboratory measurements of CO fumes to mine conditions- -this being due to postdetonation oxidation of CO to co2; and (3) the detonation velocity decay rate of an explosive is a useful experimental parameter for correlating toxic fumes production with nonideal detonation behavior.