Purpose Using the threshold concepts framework, this paper aims to explore how differences in the ability to meaningfully apply relevant literature to one's research are reflected in descriptions of graduate training undertaken in an academic year. Design/methodology/approach This paper used a sequential-explanatory mixed method design. Phase I analysis used quantitiatve performance data to differentiate research skill threshold crossers from non-crossers. Phase II analysis used qualitative interview data to identify common and differentiating themes across and between the two groups. Findings Participants identified coursework, research activities and teaching assignments as primary research skill development sites. However, only the patterns of mentorship and engagement with literature within the context of supervised research activities consistently differentiatied threshold crossers from non-crossers. All non-crossers reported having full autonomy in their research endeavors, whereas all crossers articulated reliance on superising mentor guidance. SImilarly, most non-crossers did not frame research as incremental contributions to existing literature, while most crossers did. Research limitations/implications The study sample size is small (n = 14), and the study is exploratory in nature. Practical implications The importance of exploring the factors that actually indicate and lead to research skill development is highlighted. Originality/value Few studies address graduate student research skill development, although this skill development is a core goal of many graduate programs. This study does so, using performance rather than self-reported data.