Most research in reading comprehension has focused on answering questions on individual documents or even single paragraphs. We introduce a neural which integrates and reasons relying on information spread within and across multiple documents. We frame it as an inference problem on graph. Mentions of entities are nodes of this graph while edges encode between different mentions (e. g. , within- and cross-document-reference). Graph convolutional networks (GCNs) are applied to these graphs trained to perform multi-step reasoning. Our Entity-GCN method is scalable compact, and it achieves state-of-the-art results on a multi-document answering dataset, WikiHop (Welbl et al. , 2018).