Purpose: The authors investigated continuity and discontinuity of vocabulary skills in a population-based cohort in the Netherlands. Method: Mothers of 3,759 children completed the Dutch version of the MacArthur Short Form Vocabulary Checklist (Zink & Lejaegere, 2003) at 18 months and a Dutch translation of the Language Development Survey (Rescorla, 1989) at 30 months. At both ages, expressive vocabulary delay was defined as vocabulary scores less than 10th age- and gender-specific percentile. Results: Of the children, 85.2% had normal vocabulary development at both ages, 6.2% were "late bloomers," 6.0% had late onset expressive vocabulary delay, and 2.6% had persistent expressive vocabulary delay. Word production and comprehension at 18 months explained 11.5% of the variance in 30-month vocabulary scores, with low birth weight, child age, gender and ethnicity, maternal age and education, and parenting stress explaining an additional 6.2%. Multinomial logistic regression was used to identify biological, demographic, and psychological factors associated with each of the vocabulary delay outcome groups relative to the typically developing group. Conclusions: Although multiple perinatal, demographic, and maternal psychosocial factors significantly predicted vocabulary skills at 30 months, positive predictive value and sensitivity were low. Future studies should address to what extent additional factors, such as brain maturation and genetic influences, can improve the prediction and understanding of continuity and discontinuity of language delay.