AIM: This study re-evaluated the classification accuracy of language screening. This is because the new Swedish child health programme moved this screening from 3 to 2.5 years of age. Another aim was to investigate the stability of diagnoses between these two time points. METHODS: Children were recruited consecutively during 2016-2017 from three Child Health Services in Gävle, Sweden. Forty-eight monolingual children (31 boys) and 93 bilingual children (38 boys) underwent language screening and a clinical examination by a speech and language pathologist at 2.5 years of age. They were re-examined after 6 months (at 36-38 months) for clinical diagnosis. RESULTS: Of the 48 monolingual children, 45 retained their status at both 2.5 and 3 years of age, while three no longer met the criteria for development language disorder. Among the 93 bilingual children, 87 retained their 2.5-year status at age 3, two no longer met the criteria, and four new cases were diagnosed. These differences were not statistically significant. All changes in screening parameters between 2.5 and 3 years were within the 95% Confidence Interval, indicating stable classification. CONCLUSION: Screening accuracy and language status were robust between 2.5 and 3 years, providing no support for a wait-and-see approach.