Particle size distributions measured by the Engine Exhaust Particle Sizer Spectrometer (EEPS) have been reported to disagree with those by scanning mobility particle sizers (SMPS). The discrepancies are larger for the accumulation mode engine exhaust particles than for compact-shape particles. Engine exhaust particles, specifically carbonaceous aggregates, have different charging characteristics compared to nearly spherical particles because aggregates acquire more charges at a given mobility diameter in a unipolar charging environment. Therefore, different instrument matrices, that represent the relationship between particle size and EEPS electrometer current distributions, are needed for compact-shape particles and aggregates. This paper reports on a study to improve the EEPS performance for engine exhaust measurements. The EEPS was calibrated with monodisperse particles from the exhaust of a diesel engine. The geometric mean diameters measured by the EEPS using the original instrument matrix agreed with SMPS within 15% for particles <50 nm (that were mostly spherical), but underestimated by 20-50% for larger particles (that were mostly aggregates). A new EEPS instrument matrix was developed for fresh engine exhaust, and its performance was evaluated using a diesel engine, two passenger cars with gasoline direct injection (GDI) engines, one vehicle with a turbo direct injection (TDI) diesel engine, and a diesel generator. The geometric mean diameters by the EEPS new matrix agreed with SMPS within 14% for monodisperse engine exhaust particles <50 nm while larger particles agreed within 19% to 1%. The new matrix also improved the agreement between EEPS and SMPS for polydisperse exhaust particles from different engines under different operating conditions. Diesel engine tests showed that the total concentration, geometric mean diameter, and geometric standard deviation by the SMPS and EEPS with the new matrix differed less than 16%, 33%, and 9%, respectively, with the greatest differences found for particles <15 nm. (C) 2015 Elsevier Ltd. All rights reserved.