Question Answering (QA) systems take a natural language question from the user, and return the precise answer from the set of documents. QA systems in the medical field can bring significant time saving to the patients, and others that need professional advice. The existing Amharic QA systems handle fact and non–fact based questions types separately for open domain. However, this type of QA systems does not support medical question and people who have difficulty using their hands to access information. Therefore, to address these problems speech enabled medical question answering is needed. This study develop speech enabled Amharic medical question answering system for both fact and non-fact question types. Towards this end, literatures are reviewed and tools such as Lucene tool for question answering, Sphinx tool for speech recognition and NetBeans to develop the prototype are used. The Amharic speech-based medical question answering system accepts speech medical queries (question) from the user and converts to text using the Amharic speech recognition subsystem. Then the Amharic medical question answering subsystem takes the text output of the Amharic Speech Recognition to extract the correct answer from the corpus. Finally, extracted answers are displayed to the users in the form of text. Each subsystems are evaluated and experimental result shows that the performance of Amharic speech recognition registered 11.45% WER during development testing and 80.93% recognition performance using microphone speech input data. The performance of question answering is 74% average precision in retrieving correct answers. After integrating, the performance registered using speech-based Amharic medical question answering system is 69 % average precision in retrieving correct answer. Nevertheless, the performance of speech based medical question answering system is greatly affected by the speech recognition subsystem system, number for documents in the corpus and answer extraction techniques applied.