In a recent study of conceptual combination, Estes (2003) presented evidence for the priming of relational information in the absence of shared constituents between the prime and target (e.g., "pancake spatula" was interpreted more quickly following "bacon tongs" than following "city riots"). He argued that these data support the view that relations exist as independent conceptual structures rather than being tied to the constituent concepts. This conclusion might be premature. Experiments 1 and 2 show that both the modifier and head noun constituents in the target and prime were much more semantically similar in the same relation condition than in the different relation condition (e.g., "pancake" and "bacon" vs. "pancake" and "city"; "spatula" and "tongs" vs. "spatula" and "riots") suggesting that the relation priming found by Estes was, in fact, semantic priming. Experiments 3 and 4 demonstrate that when the semantic similarity of the constituent words is controlled, there is no relation priming in the absence of repeated constituents. These results are discussed in light of previous literature suggesting that the relational information used in conceptual combination is tied to the constituent concepts, particularly the modifier.