Explainable Graph Neural Network (GNN) has emerged recently to foster the trust of using GNNs. Various GNN explainers from different perspectives have been designed. Alternatively, we advocate it is also crucial to study the robustness of GNN explainers in the face of adversaries, due to the security implications in real-world applications. However, this is unexplored. We take the first step on exploring GNN explainers under adversarial attack—An adversary slightly perturbs graph structure such that a GNN model makes correct predictions, but the GNN explainer yields a drastically different explanation on the perturbed graph. Specifically, we first formulate the attack problem under a practical threat model (i.e., the adversary has limited knowledge about the explainer and has restricted perturbation budget). We then design two methods (i.e., one is loss-based and the other is deduction-based) to realize the attack. We evaluate our attacks on various GNN explainers and datasets, and the results show these explainers are fragile. We hence call for designing robust GNN explainers in future work.