Current mathematics curricula and assessments emphasize students' ability to communicate mathematical reasoning in writing. Limited research exists, however, on what mathematics writing (MW) is and how to assess the quality of MW. Therefore, based on our definition of MW construct, we explored 4 approaches to scoring MW: (a) holistic rubric scoring for overall quality of explanation; (b) analytical rubric scoring across 4 components (i.e., mathematics content, mathematics vocabulary, writing organization, writing grammar); (c) genre-based elements scoring (modified from the scoring form used for the Essay Composition subtest of the Wechsler Individualized Achievement Test); and (d) curriculum-based measurement mathematics-writing sequences (MWS) scoring, including rules for scoring numbers and symbols. Students from 6 classrooms in 3rd-grade, 4th grade, and 5th grade (2 classrooms at each grade level; N = 122) were recruited for the current study. Results indicated all 4 scoring methods were moderately correlated with measures of writing and mathematics. Holistic, analytic, and elements scoring were strongly correlated with each other, with slightly lower correlations for the MWS scoring. Confirmatory factor analysis further demonstrated adequate construct validity. All 4 scoring methods were reliable with an exception of lower reliability for writing components assessed by analytic scoring. Based on the validity, reliability, efficiency, and information provided by each scoring method, we recommend holistic scoring for summative assessments and elements scoring for formative assessments.