Second-language (L2) research, including L2 writing research, typically recruits college or university students. There is an increasing awareness that this sampling bias entails negative consequences both for the generalizability of emerging L2 theories and for the applicability of research findings in heterogeneous L2 classrooms. However, previous research has not systematically investigated the association between adults’ L2 writing development and their prior level of formal education. The present study contributes a concrete portrayal of what that association may look like by investigating a text corpus collected longitudinally from 38 students enrolled in Swedish for Immigrants. The texts were assessed by five experienced L2 teachers using the method of comparative judgement, which ascribes a holistic quality score to each text. The participants follow three study paths that reflect their varying levels of education, and statistical analyses revealed that study path significantly predicted the participants’ writing ability and their rate of writing development. Despite the small sample size and other limitations of the study, those results confirm the importance of treating educational background as a crucial factor in L2 writing research. Implications for curriculum design are discussed, and methodological challenges that further L2 research needs to address are highlighted.