Two studies test the hypotheses that individual differences in phonological memory among children younger than two years can be assessed using a non-word repetition task (NWR) and that these differences are related to the children's rates of vocabulary development. NWR accuracy, real word repetition accuracy and productive vocabulary were assessed in 15 children between 1 ; 9 and 2 ; 0 in Study 1 and in 21 children between 1 ; 8 and 2 ; 0 in Study 2. In both studies, NWR accuracy was significantly related to vocabulary percentile and, furthermore, uniquely accounted for a substantial portion of the variance in vocabulary when real word repetition accuracy was held constant. The findings establish NWR as a valid measure of phonological memory in very young children, and they open the door for further studies of the role of phonological memory in early word learning.