Nanotechnology has been presented in the policy discourse as an intrinsically interdisciplinary field, requiring collaborations among researchers with different backgrounds, and specific funding schemes supporting knowledge-integration activities. Early bibliometric studies supported this interdisciplinary vision (MEYER & PERSSON, 1998), but recent results suggest that nanotechnology is (yet) a mixed bag with various mono-disciplinary subfields (SCHUMMER, 2004). We have reexamined the issue at the research project level, carrying out five case studies in molecular motors, a specialty of bionanotechnology. Relying both in data from interviews and bibliometric indicators, we have developed a multidimensional analysis (SANZ-MENÉNDEZ et al., 2001) in order to explore the extent and types of cross-disciplinary practices in each project. We have found that there is a consistent high degree of cross-disciplinarity in the cognitive practices of research (i.e., use of references and instrumentalities) but a more erratic and narrower degree in the social dimensions (i.e., affiliation and researchers’ background). This suggests that cross-disciplinarity is an eminently epistemic characteristic and that bibliometric indicators based on citations and references capture more accurately the generation of cross-disciplinary knowledge than approaches tracking co-authors’ disciplinary affiliations. In the light of these findings we raise the question whether policies focusing on formal collaborations between laboratories are the most appropriate to facilitate cross-disciplinary knowledge acquisition and generation.