This study combines research methods from student approaches to learning research and social network analysis (SNA) to examine patterns of students' collaborative learning based on their learning orientations amongst 193 postgraduates enrolled in a blended course. The study identified two distinct learning orientations, namely 'understanding' and 'reproducing', which differed in approaches to learning through inquiry, approaches to using online learning technologies, perceptions of the online workload, and academic outcomes. On the basis of students' learning orientations and their choice of whether to collaborate and with whom to collaborate, five networks representing five patterns of collaborative learning were found. From these, two did not reveal any collaboration (Understanding Alone and Reproducing Alone networks); and three revealed collaborations (Understanding Collaboration, Mixed Collaboration, Reproducing Collaboration networks). A range of SNA measures were calculated and revealed different features of the three collaboration networks. Viewed together, the combined methodologies suggest that the Understanding Collaboration network has more desirable features of collaboration, such as the intensity of collaboration, having closely knitted groups who tended to seek out and welcome peers and who tended to engage more often in both face-to-face and online modes. The study suggests that helping students adjust their learning orientations, designing some compulsory collaborative assessment tasks, and configuring the composition of collaborative groups are productive strategies likely to improve students' experiences of collaborative learning.