The effects of two successive annual treatments of imidacloprid and fipronil on dragonfly nymph communities, which are one of the best-known bioindicators in Japanese agroecosystems, were monitored in experimental paddies. The abundance of dragonfly nymphs was lower in both insecticides-treated fields than it was in the controls, particularly following fipronil treatments. Residues of both insecticides were found in the soil throughout the two years, and imidacloprid persisted in water up to three months following each treatment. A Principal Response Curve analysis (PRC) showed that the second annual treatments caused greater structural changes in dragonfly nymph communities than the initial treatments caused, particularly for fipronil. The community structures continued to change even after the insecticides dissipated from the water. This suggests that ecological impacts, and therefore risks, of imidacloprid and fipronil on dragonfly nymph communities depend more on soil residues than they do on waterborne residues. As expected, susceptibility of dragonfly nymphs to these two insecticides differed among species.