This study of Japanese New Product Development (NPD) projects explores the fuzzy front end of innovation. The goal of the paper is twofold: First, we analyse the impact of the fuzzy front end on the success of all the NPD projects. We develop a conceptual model based on the information-processing approach. A structural equation model was fitted to data from 497 NPD projects in Japanese manufacturing firms to test the proposed model. The empirical analysis suggests that an early reduction of market and technical uncertainty as well as an initial planning before development have a positive impact on NPD project success. The model accounts for 17% of the variance of the efficiency and 24% of the variance of the effectiveness-dependent variable. Thus, the front end phase is an important driver of NPD project success. Second, we compare the fuzzy front end of incremental to radical NPD projects. Although these projects differ in many aspects of newness, we found only a few differences with regard to the fuzzy front end. For instance, it was more difficult to estimate the market size and price sensitivity of the customers during the fuzzy front end of the radical NPD projects compared with the incremental projects. Implications of the empirical results and limitations of the study are discussed.