Background: Stroke is the second leading cause of death worldwide, with high morbidity and disability rates. In recent years, some scholars have used commonly nutrition screening and evaluation tools to screen and evaluate the nutritional status of patients with stroke. Some scholars have also selected some subjective nutritional evaluation scales to study the prognosis of stroke. However, there is a lack of relevant systems to evaluate the association of different nutritional screening and assessment scales with stroke outcomes.Methods: Pubmed, Embass, Web of science, and Cochrane databases were searched to induce cross-sectional studies of stroke outcomes using nutritional assessment scales. The main indicators and secondary indicators were extracted for meta-subgroup analysis.Results: Thirty articles were included in this study, involving 11604 patients from 14 regions. Two types of data analysis were performed on NRS-2002, MUST, MNA, MNA-SF, SGA, and PG-SGA to predict and evaluate the prognosis of stroke, among which NRS-2002 was 0.44 [95% CI: 0.37–0.52], MUST was 0.58 [95% CI: 0.45–0.70], MNA was 0.29 [95% CI: 0.10–0.52], MNA-SF was 0.11 [95% CI: 0.00–0.42], and SGA and PG-SGA were 0.22 [95% CI: 0.00–0.71]. Continuous data analysis of the mNUTRIC, NRS-2002, and SGA scales used for analysis revealed a value of 13.059 [95% CI: 8.856–17.263].Conclusion: MNA and MNA-SF are valuable in predicting life activities after stroke in elderly patients and evaluating the stage of recovery after stroke. However, mNUTRIC scale is more effective in the assessment of acute stroke. The results of MUST and NRS-2002 provided a basis for early nutritional intervention. SGA and PG-SGA can be continuously to evaluate nutritional interventions after stroke.