In the knowledge-based economy, innovation is seen to play a central role, but until recently the complex processes of innovation have been insufficiently understood. Better understanding, however, has emerged from many studies in recent years. At the macro-level, there is a substantial body of evidence that innovation is the dominant factor in national economic growth and international patterns of trade. At the micro-level, R&D is seen as enhancing a firm’s capacity to absorb and make use of new knowledge, strengthening its competitiveness. As innovation comes into a focus, more concerns have been paid to how to collect and produce indicators which well represents the economic activities in general and S&T activities in particular. The attempt to seek and develop new indicators is inevitable since conventional indicators are not quite appropriate for capturing innovation activities and therefore for policy-making for various socio-economic sectors. Conventionally, there have been two basic families of S&T indicators, which are directly relevant to the measurement of technological innovation, that is, resources devoted to R&D and patents statistics. R&D data are collected through national surveys according to the guidelines of the Frascati manual. These data have proved valuable in many studies. However, these data have two main limitations. First, R&D is an input. Although it is obviously related to technical change, it does not measure it. Second, R&D does not encompass all the efforts of firms and governments in this area, as there are other sources of technical change, such as learning by doing. On the other hand, patents as an indicator also have a drawback. Many innovations do not correspond to a patented invention. Many patents correspond to invention with a near zero technological and economic values, whereas a few of them have very high value. Many patents never lead to innovation. Thus, it is needed to develop new indicators properly reflecting innovation activities. In this study, we have made an investigation of firm’s innovation behavior and provided some empirical evidences, based on the KIS (Korean Innovation Survey) dataset, which is collected according to the OECD scheme. In the following chapter, we will discuss the empirical model and data mining. The empirical study was done employing variables, which explains firm’s innovation behavior. Those variables include not only conventional variables such as firm’s age, size and industrial structure, but also the variables such as technological opportunities, in-house capability and market environment, which are perceived by the firm. Using those variables, we obtained some empirical evidences using a probit model. An advantage using an econometric model is that we can systematically reproduce the results at given conditions. Then, we provide some explanation of the empirical results and finally concluding remarks in the last chapter.