Previous research has shown that innovation can have various sources and different forms. Innovation can thus be developed by different types of organizations or individuals for several reasons, and it can be related to product or process technologies. One particular stream of research has explored the role of users as a source of innovation. Examples of user innovations can be found in a variety of fields and settings such as mountain biking, snowboarding, open source software, scientific instruments, oil refining, semiconductor manufacturing, and even the World Wide Web. What these innovations have in common is that they were all initially developed by people or organizations that wanted to solve a specific need and benefited from using their innovation rather than selling it. A particular type of user innovator is a "user firm" which develops new or improved production technology for its own internal use. Such process innovation is characterized by determinants and outcomes, which are fundamentally different from for example product innovation. In particular, process innovation may be driven by learning-by-doing, which is a form of problem-solving or experimentation that takes place on the production floor rather than in research and development (R&D). However, the exact drivers and consequences of process innovation in general and learning-by-doing in particular are not yet fully explored. Therefore, the objective of this thesis is to increase the understanding of the antecedents and impact of process innovation in user firms by exploring the role of non-R&D activities and learning-by-doing. This thesis consists of four parts. Each of them addresses a specific aspect related to the sources of process innovation in user firms. The first paper particularly argues that process innovation relies on different learning mechanisms than product innovation. Using data from the Swiss Innovation Survey of the Swiss Institute of Business Cycle Research (KOF) at ETH Zürich, the paper shows which are the external and internal knowledge sources (related to R&D, manufacturing and marketing) that lead to innovation. The second paper further investigates R&D and non-R&D activities as sources of innovation and develops two measures to quantify the magnitude of non-R&D innovation. The results show that a substantial part of the firms develop innovations without R&D and that non-R&D process innovation has a very large impact on the overall cost reductions in the Swiss economy. In order to better understand the sources and attributes of process innovation in user firms, a questionnaire was conducted in a sample of Swiss manufacturing firms. Based on the results, the third paper shows the pervasiveness of major and especially minor process innovation. It also explores the sources of process innovation within the firms and identifies the practices related to the accounting, protection, appropriation, and monitoring of process innovation, which are often of an informal nature. The fourth paper investigates in more detail how such practices drive learning-by-doing and process innovation. The findings firstly show which complementary systems of practices are implemented in user firms to promote the innovative contribution of production floor workers and secondly how these practices drive learning-by-doing for either major or minor process innovation.