Early identification of students at risk for writing disabilities is an important step in improving writing performance. Kindergarten students ( n = 84) were administered a set of researcher-developed writing tasks (letter writing, sound spelling, word spelling, and sentence writing) and school-administered reading tasks ( Dynamic Indicators of Early Literacy Skills [DIBELS], Phoneme Segmentation Fluency [PSF], Nonsense Word Fluency [NWF], and Letter Name Fluency [LNF] subtests [DIBELS]) in January. The students were identified as at risk based on a norm-referenced writing assessment and teacher ratings collected in the spring. The classification accuracy of the writing and reading tasks was estimated using receiver operating characteristic (ROC) curves. For both risk criteria, individual reading and writing assessments demonstrated comparable accuracy (area under the curve [AUC] statistics range = .57–.87). However, classification accuracy was strengthened when reading and writing measures were combined (AUC range = .75–.92). The results suggest that the most accurate approach to universal screening of writing difficulties may include a battery of reading and writing measures.