We carried out three experimental studies to investigate the upper critical temperatures of cereal aphids, Rhopalosiphum padi, Sitobion avenae and Schizaphis graminum, for inhabiting, crawling and feeding with self-made temperature gradient apparatus. The temperature gradient for testing inhabiting and crawling of aphids was generated across a wheat leaf which floated on the water in an enamelware tray by heating one side of the tray with an incandescent light under the tray. The gradually increasing temperature for testing feeding behavior of aphids was generated on a wheat leaf which floated on the water in a plastic petri dish by setting an incandescent light under the petri dish for heating. Temperatures of the two kinds of temperature gradients were controlled by lengthening or shortening the spatial distance between the top of the adjustable incandescent light and the bottom of enamelware tray or plastic petri dish. Infrared thermometer was used to measure the temperatures on wheat leaves in each study. Digital camera was used to record the distribution of tested aphids in leaf temperature gradient when studying the upper critical temperatures of cereal aphids for inhabiting. Upper critical temperature for inhabiting is defined as the temperature below which the accumulative percentage (P_a) of tested aphids distributed in certain temperature gradient had reached to 90%. Our results indicate that, in the leaf temperature gradient (26-43℃), the upper critical temperatures of R. padi, S. avenae and S. graminum for inhabiting were 30.2, 28.8℃ and 27.3℃ respectively, when the Pareached to 90%. P_a were influenced by the temperatures along the temperature gradient and can be described with a “Sigmoidal” model. Upper critical temperature for crawling is defined as the temperature at which the tested aphids began to turn back when they were moving along the temperature gradient from the cool side to the hot side. Meanwhile, upper critical temperature for feeding is defined as the temperature at which the tested aphids began to pull out the stylet from the host leaf and escape, when they were feeding on the increasingly heated plant leaves. Our results show that the upper critical temperatures of R. padi, S. avenae and S. graminum were 42.0, 39.1℃ and 38.5℃ respectively, for crawling; and 39.3, 40.2℃ and 39.0℃ for feeding. Our results show that R. padi has higher tolerance to high temperature than S. avenae and S. graminum. We discussed the potential effects of temperature extent, relative humidity and source of the tested aphid populations on the behavioral response of cereal aphids to temperature gradient. More detailed investigation should be carried out in the future on favourable temperatures for ecdysis of nymphs and reproduction of adults, and the differences between apterous and alatae for temperature preference. Our results may be helpful to improve the sampling method of the cereal aphids in the fields, and enhance the accuracy of forecasting the infestation of cereal aphids. Our experimental apparatus and methods supply an example in studies on temperature preference of other species of insects.