Query Reformulation (QR) is a technique that transforms user queries to better align with their intent, enhancing search experience. Zero-shot QR leverages large language models (LLMs) to generate query reformulations without labeled examples. This paper introduces GenQREnsemble, an ensemble-based prompting technique that uses multiple paraphrases of a zero-shot instruction to generate diverse sets of keywords, improving retrieval performance. The post-retrieval variant, GenQREnsembleRF, incorporates pseudo-relevance feedback to further enhance performance. Evaluations on four IR benchmarks show that GenQREnsemble improves nDCG@10 by up to 18% and MAP by up to 24% compared to previous zero-shot state-of-the-art methods. GenQREnsembleRF also demonstrates gains of 5% MRR and 9% nDCG@10 on the MSMarco Passage Ranking task using pseudo-relevance and relevant feedback, respectively. The paper highlights the benefits of ensemble strategies in improving query reformulation and discusses the effectiveness of incorporating relevance feedback.Query Reformulation (QR) is a technique that transforms user queries to better align with their intent, enhancing search experience. Zero-shot QR leverages large language models (LLMs) to generate query reformulations without labeled examples. This paper introduces GenQREnsemble, an ensemble-based prompting technique that uses multiple paraphrases of a zero-shot instruction to generate diverse sets of keywords, improving retrieval performance. The post-retrieval variant, GenQREnsembleRF, incorporates pseudo-relevance feedback to further enhance performance. Evaluations on four IR benchmarks show that GenQREnsemble improves nDCG@10 by up to 18% and MAP by up to 24% compared to previous zero-shot state-of-the-art methods. GenQREnsembleRF also demonstrates gains of 5% MRR and 9% nDCG@10 on the MSMarco Passage Ranking task using pseudo-relevance and relevant feedback, respectively. The paper highlights the benefits of ensemble strategies in improving query reformulation and discusses the effectiveness of incorporating relevance feedback.