Abstract: There are two locational patterns of R&D's investment.One is concentrating and the other is dispersing spatially. The Schumpeterian strand of the endogenous growth approach has highlighted the advantages of spatially concentrating the R&D effort in a few areas, in order to maximize external economies and technological spillovers. Innovation is then expected to spill over from these technologically advanced areas to neighboring regions. The neoclassical view, in contrast, considers that decreasing returns means investment in core areas is less effective than that in peripheries. Based on the empirical test of the locational patterns of R&D's investment in China, from the point of investment and effect of output, we find that it has more advantages of spatially concentrating R&D's investment in technologically advanced areas. From the views of the intensity of R&D's investment, and the promoting effect on economic development, scientific research, industrialization, there have six types of areas which are distinct from each others. In the six types of areas, there are the scientific pattern which are often belong to few technologically advanced areas, and the industrial pattern which can use for reference to technologically lagging areas. At last, the comparison and analysis of the gaps of technologically lagging areas have been conducted.