In cross-situational word learning, individuals associate words with their meanings by observing their use across multiple encounters. This is considered to be a crucial mechanism involved in early childhood language acquisition. However, whether learners can track multiple words for referents cross-situationally, such as when learning synonyms, remains poorly understood. The present study investigated the effects of age on children’s cross-situational word learning (CSWL) from a two-to-one structure, where objects are given two names. Younger (4 to 5.4 years, N = 29) and older (5.5 to 7.9 years, N = 48) children completed a cross-situational word learning task in which, during the first half of training, objects were labeled with one label (First words) and in the second half of training, objects were given a new second label (Second words). Results showed that age interacted with learning: younger children learned second labels but not first labels, whereas older children learned first labels but not second labels. These findings indicate the limitations of children’s capacity to learn complex word-referent mappings in CSWL.