Two experiments examined how context and syntactic priming interact to determine interlocutors’ choice of referential form. Pairs of naïve participants took turns producing descriptions of target pictures from a set of alternatives. The first experiment established that a contrast picture in the display (e.g. a striped cat in a display where a spotted cat was the target) primarily determined whether an adjective was used. Priming with an adjective had a only small, secondary effect on adjective use. However, when an adjective was used, it was more likely to occur in the prime-congruent structure than the alternative structure. Experiment 2 compared the effects of a prime produced by the dialogue partner with the effects of a pre-recorded prime played through headphones. Syntactic priming was significant only for the dialogue prime trials, indicating that priming may be stronger in dialogue than outside of dialogue, as previous work has suggested. However, even in dialogue, the primary factor that determined referential form was the set of alternatives. Our results begin to clarify the role of syntactic priming in dialogue, suggesting that it has at most a small effect on message formulation.