AIM\nPrevious findings regarding the prevalence and predictive value of weak language skills in preterm children with very low birth weight (VLBW) are unclear. This study analysed the prevalence of weak language skills, the predictive value of early weak language skills on later weak language skills, and the sensitivity and specificity of cognitive scores for identifying concurrent weak language skills in a longitudinal sample of VLBW children (n = 141) and their full-term controls (n = 146).\nMETHODS\nData on language skills and cognitive development were gathered at two and five years of age. Weak language skills were defined by the 10th percentile value of the controls.\nRESULTS\nIn VLBW children, the prevalence of weak language skills varied between 16% and 18% at 2 years of age (controls: 8 to 10%) and between 20% and 27% at 5 years of age (controls: 10%). Early weak language skills predicted later weak language skills in VLBW children. Cognitive scores were specific, but their sensitivity for identifying concurrent weak language skills was low.\nCONCLUSION\nThe prevalence of weak language skills in VLBW children increased during the follow-up period and was higher than the controls. Language-sensitive methods should be used in the clinical follow-up of VLBW children.