A multiple linear regression technique was used to evaluate and correct the matrix interferences in the determination of As and Pb concentrations in fly ashes by inductively coupled plasma optical emission spectrometry. The direct determination of As and Pb in SRM 1633b by ICP-OES failed to obtain the certified concentrations, except in a couple of cases. However, it proved possible to use the multiple linear regression (MLR) technique to correct the determined concentrations to a satisfactory level. This method of correction is based on the multiple regression line obtained from the analysis of 19 synthetic mixtures of matrix and analyte elements (Al, As, Ca, Fe, Pb, and Si) at five concentration levels. The matrix interferences in the determination of As were caused by Al, Pb, and Ca whereas the matrix interferences in the determination of Pb were caused by Al and Fe. The most suitable parameters for the determination of As and Pb were a plasma power of 1500 W and a nebulizer flow of 0.5 or 0.6 L min−1. The accuracy of the method was shown with the analysis of SRM 1633b and two fly ash samples with the standard addition method. A recovery rate of 96% can be reached for Pb at 220.353 nm with three digestion methods (US-TSD, US1 and MW) by using both direct measurement with thoroughly optimized plasma conditions and the MLR method. A recovery rate of 93% was obtained for As when using the MLR method at 193.696 nm with the digestion method US2, a plasma power of 1500 W, and nebulizer flow of 0.6 L min−1. The corrected and determined concentrations of As and Pb in samples analyzed resulted in a precision of 0.6 to 3.9%.