This paper reports on a quasi-experimental research investigating the effects of using a writing assistance software (WAS) on the students’ quality of writing. Three hundred and four teenagers enrolled in French public schools in Quebec participated in this study. Students wrote a first text by hand, and then a second text under one of the following modalities: 1) handwriting; 2) word processor (WP) only; 3) WP and WAS; 4) WP, a WAS and prior training. Data has been analyzed from a quantitative perspective (ANOVAs, paired t-tests). Findings showed that WAS users made less grammar and spelling mistakes than their counterparts did. However, they tended to make more vocabulary mistakes. On the other hand, students using only WP did less well with digital tool than by hand. Overall, these findings emphasize that learning to write in a digital environment is a complex process: such complexity should be taken into account in schools.