This study aims at gaining more insight in effective writing instruction to promote historical reasoning. In an experimental study, two types of instructions were compared; a general writing instruction and a discipline-based writing instruction. In addition, the effects of these instructions for students with a different initial writing ability were explored. Participants were 42 students (11th-grade), who followed a unit on the development of Dutch democracy and had to write an argumentative letter in which they argued the historical significance of a self-selected event or person. Students received a short writing instruction, based on the principle of learning from text models, in two versions: a general writing instruction or a discipline-based writing instruction. Analyses focused on historical reasoning and global text quality. Results showed a positive effect of discipline-based instruction on the quality of historical reasoning, but no effects were found on text quality. No differences were observed for good and weak writers. A pre- and post knowledge test showed improvement of students' knowledge, but no differences between conditions were found. The outcomes add to earlier studies that found positive effects of discipline-based writing instruction and provide teachers with directions for designing discipline-based writing instructions based on learning from text-models.