Vocalizations of titi monkeys were recorded from 6 groups at two sites in the llanos of Colombia during 8 months of study. Frequency and temporal measurements from spectrograms of recordings, and differences in their patterns of use, were used to characterize call types. Animals repeat calls to form phrases, and combine phrases to form sequences. The six sequence types, defined by different transition probabilities between phrases, varied in proportion in different contexts. To examine the effect of order of phrases on response, I played back artificially constructed sequences to subject groups. I discuss hierarchical and sequential mechanisms as generators' of vocal sequences, and draw comparisons with models of human language.