This validation study was undertaken to evaluate the construct of Formative Assessment of Writing (FAoW) operationalized by an instrument with 50 Likert scale items. To identify the EFL learners' experiences of FAoW practices, the instrument was first piloted with three EFL learners, and subsequently administered on a sample of 255 EFL learners selected based on purposive sampling. A five-factor solution with five latent variables (i.e. clarifying criteria, evidence on students' current learning, feedback to move learners forward, peer-assessment and autonomy) was evaluated through Confirmatory Factor Analysis (CFA) with AMOS 22. Model fit showed that the five-factor structure of FAoW could only be supported in terms of absolute and parsimony fit indices. The model with three factors (i.e. clarifying criteria, peer-assessment and feedback) in two stages of pre- and while-writing, however, provided higher discriminant validity in addition to absolute and parsimony fit indices. In other words, FAoW was not found to be practiced within its full potential with five components in the context of this study. A conceptual model was developed based on the findings and the literature to show pedagogical application of FAoW and how it can be practiced in line with Black and Wiliam's (2009).