Doctoral studies are about learning to create new knowledge and to become a researcher. Yet surprisingly little is known about the individual learning patterns of doctoral students. The study aims to explore learning patterns among natural science doctoral students. The participants included 19 doctoral students from a top-level natural science research community. The data were collected through interviews and qualitatively content analysed. Five qualitatively different learning patterns were identified: 1) active knowledge creator, 2) active producer, 3) active project manager, 4) passive producer, and 5) conformist. The patterns differed from each other in how the participants approached their learning regarding conducting research and becoming a researcher, learning strategies and their perceptions of learning objects. This indicates that learning environments need to be adjustable to different doctoral student learning patterns. At best, by designing practices in congruence with doctoral students' ways of learning, scholarly communities can provide flourishing environments in which doctoral students are able to become autonomous scientists who conduct high quality research. To our knowledge, doctoral students' learning patterns have not been previously reported in doctoral education literature. This study contributed to the literature on doctoral student learning and provided new insight into the complexity of learning processes among natural science doctoral students by identifying five qualitative different learning patterns.