This study aimed to investigate the predictability of writing development and if scores on a writing test in the first weeks of first grade accurately predict students' placements into different proficiency groups. Participants were 832 first grade students in Norway. Writing proficiency was measured twice, at the start and at the end of first grade (time 1 and time 2, respectively). Multilevel linear regression analysis showed that writing proficiency measures at time 1 were significant predictors of writing proficiency at time 2. The results also showed that measures at time 1 could identify students running the risk of not meeting expectations with high precision. However, the results also revealed a substantial proportion of false positives. The results are interpreted and discussed from a formative writing assessment perspective.