The situation of industrial technology development in Taiwan is somewhat different from other countries. The most distinctive feature of technology and science policies is that the technology development of strategic industries largely dependent on those research institutes, for example, Industrial Technology Research Institute, which expends couple of billion NT dollar budget from government and occupy the biggest R&D pie in public section. Such institutes receive budget mainly from government but partially from private enterprises, and their goals are cresting new technologies for transferring. The protection of intellectual property right became critical competence in radical innovation industries. In addition to patent style protection, trade secrets provide another way to protect R&D results. Lawful protections not only resist competitors, but also promote competitive edge. For example, Japanese intensively apply patents in industries of semiconductor, mechanics, and optical electronics promote national competence in those industries. In order to achieve an objective of optimization of resources application, the decision of choosing protection style must correlate all substantial factors. There are some fragmentary and piecemeal studies discuss such issue, the systematically theoretical and empirical study of this issue still be waiting, especially with managerial viewpoint. The majority of studies on protection of intellectual property focuses on single style, either patent or trade secret. The purpose of this paper distinguishes from prior articles on two points, one is the target of research institutes, and another is processing empirically. In industry technology development, patent and trade secret are two similar ways to protect the same matter, but differently while consider the related influential factors. The characteristics of patent and trade secret and their relationship were discussed in literature review. Five factors were found which were the most correlated factors that influence the decision of choosing protection mode for R&D results between patent and trade secret in research institutes. Therefore, five hypotheses were followed to these five factors. H1: The more clear attitude of senior executives, the more research institute trended to patent application. H2: The more possibility of licensing-out, the more research institute trended to patent application. H3: The more patentability, the more research institute trended to patent application. H4: The more possibility that the lifetime of technology was shorter than patent application period, the more research institute trended to trade secret. H5: The more the success of reverse engineering technology, the more research institute trended to trade secret. The analytic unit of this study was each decision while research institute consider the mode to protect their R&D results. The respondents should be the project leaders. Sampling was strictly restricted to the decisions that were made between 1998 and 2000 the period three years ahead of this study. According the statistics, 25 effective samples must be collected. Actually, 51 effective samples were received by non-probability sampling. Within these samples, the majority of decision came from information and communication industries, which were described as high-density of R&D activities. ANOVA and discriminative analysis were adopted for data processing. ANOVA was used to hypotheses test and discriminative analysis was used to build discriminative equation. The factors that influence the decision of research institutes to choosing the protection mode between patent and trade secret for their R&D results is the goal of this research. Based on our empirical study, attitude of senior executives, possibility of licensing-out, and reverse engineering technology are significant factors to influence such decisions.