Experimental design was used as a tool to define the optimum pyrolysis and atomization temperatures for four analytes (Cu, Pb, Ni and Cd) in biodiesel samples. Two chemical modifiers (Pd + Mg and W) and two distinct sample preparation procedures (microemulsion and wet digestion in a focused microwave system) were also investigated. The pyrolysis and atomization temperatures were optimized using 2(4) factorial design for Cu, Pb, Ni and Cd, with 16 assays carried out for each analyte. The results for Cu and Pb indicated that variables of sample preparation for digestion by focused microwave was the most important one for both analytes. The pyrolysis and atomization temperatures applied were 1000 degrees C and 2200 degrees C for Cu and 500 degrees C and 2000 degrees C for Pb. None of the variables analyzed here were important for Ni, and the pyrolysis and atomization temperatures chosen for this element were 800 degrees C and 2300 degrees C. A different factorial design was used for Cd. The variables of medium and modifier were not important for this element, and the lowest temperatures, Tp-500 degrees C and Ta-1400 degrees C, were chosen based on this second design. The importance of factorial design in the simultaneous optimization of several variables studied by GFAAS was confirmed, for it involves fewer experiments and hence, lower costs, greater speed and higher efficiency. (c) 2009 Elsevier Ltd. All rights reserved.