The purpose of the present study was to investigate knowledge-creating agency by examining doctoral students' accounts of their pursuits, using structured interviews. We examined all of the talk apparently related to agency of 13 doctoral students taking part in collective doctoral training in two, highly regarded Finnish research communities (natural science and medicine). The doctoral process involved the participants pursuing article-based theses based on collectively shared research problems and journal articles co-authored with the supervisor and other senior researchers. Based on the qualitative analysis of the interviews, three categories of agentic talk were identified, respectively, for three proposed types of "knowledge-creating agency": "personal," "distributed," and "object-related". "Personal agency" involved participants' reflection on their academic competence, self-efficacy, and personal strengths and weaknesses. "Distributed agency" played an important role in assisting these participants overcoming challenges; this includes sharing of expertise, receiving social support, and having a sense of collective efficacy. "Object-related agency," in turn, represented integration of a doctoral student's efforts with those of her immediate research community while pursuing shared research objects (e.g., co-authored articles). Apart from issues of agency which, the authors propose, are raised in the interviews, the data provide content-rich accounts of the collective nature of doctoral experiences.