Question and answer (Q&A) documents are a new type of knowledge document composed of a question part and an answer part. The questions represent knowledge needs, and the answers contain the knowledge that meets these knowledge needs. An overload of accumulated Q&A documents decreases the reuse of valuable knowledge. In this paper, we propose a novel hybrid system to recommend Q&A documents to alleviate overload. First, knowledge needs are partitioned, and current knowledge needs are identified by sequentially clustering the Q&A documents. Second, a content-based (CB) recommendation method, a collaborative filtering (CF) recommendation method and a complementarity-based recommendation method are used to find the Q&A documents that are potentially helpful for the user. Third, the three initial recommendation lists of Q&A documents derived from the three recommendation methods are combined to form a more comprehensive recommendation based on the Fermat point. Because reading all Q&A documents in the recommendation list consumes an enormous amount of time and users prefer to read Q&A documents one by one starting from the top, a novel ranking mechanism is proposed to ensure that users obtain comprehensive knowledge to the greatest extent possible from the limited number of Q&A documents at the top of the list. The proposed approach is evaluated and compared based on an experimental dataset. Our experimental results show that the approach is feasible, performs well, and provides a more effective way to recommend Q&A documents.