Peer and self-assessment methods have strong theoretical underpinnings in the literature in supporting writing from adolescents through adults and across the curriculum, but empirical research assessing the impact of each method on writing achievement outcomes is lacking, particularly for different learner achievement levels. This quasi-experimental process-outcome study of 120 first-year college writers in six course sections evaluated the impact of peer and self-assessment to determine if the methods led to gains in transferable student writing achievement, with a focus on differentiating findings for levels of learner writing at onset. The differences between pre/post writing samples from peer and self-assessment classes were compared against those from control sections. Findings show peer assessment yielded mixed results: a positive effect for low-achieving, no difference from control for mid-achieving, and negative effect for high-achieving writers at onset (p = 0.005). Self-assessment was found more effective than the control for all writer levels and more effective than peer assessment for most writers. Results indicate that all learners can benefit from self-assessment, but that, despite wide support in the theory, peer assessment's impact on writing is positive for fewer students than previously thought and may have little to no effect on transferable writing skill for others. Qualitative findings describe the complexities of teaching with each method: motivation, extensive planning requirements, strategy support, student and teacher time and energy demands. Peripheral benefits of collaboration and individualised teacher support are described.