1-Nitropyrene (1-NP), 2-NP and 2-nitrofluoranthene (2-NFR) are useful markers for studying the atmospheric behaviors of polycyclic aromatic hydrocarbons (PAHs) and nitropolycyclic aromatic hydrocarbons (NPAHs). However, present methods for measuring trace levels of these compounds are lesssensitive and laborious. Here we describe several improvements to a previously reported high-performance liquid chromatography-chemiluminescence detection system that allows it to determine trace levels of 1-, 2-NPs and 2-NFR. The proposed system was equipped with a reducer column packed with Pt/Rh instead of zinc whose life-time was limited. The combination of Cosmosil MS-II (monomeric ODS) and AR-II (polymeric ODS) columns was used instead of polymeric ODS columns as the separator column to improve the separation. An ethanol mixture with acetate buffer (pH 5.5) was used in place of an acetonitrile mixture with the same buffer to activate the reducer column. The same ethanol mixture was used as the mobile phase for the clean-up column. The switching time of the column switching valve was optimized to concentrate the amino-derivatives of above NPAHs quantitatively on the concentrator column. The concentrations of bis(2,4,6-trichlorophenly) oxalate and hydrogen peroxide in the chemiluminescence reagent solution were optimized to 0.4 mM and 30 mM, respectively, to increase the sensitivity. Under the above conditions, the detection limits (S/N=3) of 1-, 2-NPs and 2-NFR were 1 fmol (0.25 pg), 10 fmol (2.5 pg) and 4 fmol (1 pg), respectively. The proposed system was effectively used to determine trace levels of 1-, 2-NPs and 2-NFR in airborne particulates collected at Noto Peninsula. The atmospheric concentrations of 1-, 2-NPs and 2-NFR were not more than sub pg $m^{-3}$ levels. They were higher in winter (January) than in summer (July). In both seasons, the concentrations were in decreasing order, [2-NFR]>[1-NP]>[2-NP].