This paper explores the factors that affect the performance of university-industry collaborative projects, and determines whether different types of projects lead to different type of results. In particular, we examine how this performance relates to the origin, the implementation, and the financing of the collaborative project, as well as to the specific characteristics and the earlier experiences of both parties that collaborate. Our study relies on 30 in-depth, semi-structured case studies of university-industry collaborative projects. We complement that analysis with a large-scale written questionnaire sent to industrial and academic researchers in the Netherlands, examining the views of researcher towards collaborations including perceived barriers. Our results show that university-driven projects, though being more risky and troublesome, allow for unexpected fruitful scientific and technological achievements, with high spillovers to other fields. Industrial-driven projects, in contrast, result in more modest achievements, but they are more likely to be adopted for use by the participating firms. The absorption of the knowledge developed in the collaboration depends mainly on factors residing on the industrial side, such as the firm’s competences to use and further develop the knowledge in question, and the firm’s investment in knowledge transfer channels such as labour mobility. We also examined the criteria that the partners use to evaluate their collaborations. Here, we find that university and industry use quite similar evaluation criteria. Differences in their evaluation are associated with differences in the original expectations for the project. This is particularly true for publicly funded projects that aim to further develop the findings of earlier collaborations. Finally, researchers’ views on university-industry collaboration depend on their academic, entrepreneurial, and relational experiences as well as the incentives that exist in their research environment.