This paper explores the validity (sensitivity and specificity) of different cut-off levels of the UNICEF/Washington Group Child Functioning Module (CFM) and the inter-rater reliability between teachers and parents as proxy respondents, for disaggregating Fiji’s education management information system (EMIS) by disability. The method used was a cross-sectional diagnostic accuracy study comparing CFM items to standard clinical assessments for 472 primary school aged students in Fiji. Whilst previous domain-specific results showed “good” to “excellent” accuracy of the CFM domains seeing, hearing, walking and speaking, newer analysis shows only “fair” to “poor” accuracy of the cognitive domains (learning, remembering and focusing attention) and “fair” of the overall CFM (area under the Receiver Operating Characteristic curve: 0.763 parent responses, 0.786 teacher responses). Severe impairments are reported relatively evenly across CFM response categories “some difficulty”, “a lot of difficulty” and “cannot do at all”. Most moderate impairments are reported as “some difficulty”. The CFM provides a core component of data required for disaggregating Fiji’s EMIS by disability. However, choice of cut-off level and mixture of impairment severity reported across response categories are challenges. The CFM alone is not accurate enough to determine funding eligibility. For identifying children with disabilities, the CFM should be part of a broader data collection including learning and support needs data and undertaking eligibility verification visits.