Despite the rapidly increasing visibility and widespread of Machine Translation (MT), limited research has been conducted to explore the relationship between MT and English as a foreign language (EFL) learning. This paper reported the results of a survey-based study on perceptions, beliefs, and attitudes towards MT tools by EFL students. It examined their views on, experience with, and use of MT and its impacts upon language learning. A total of 68 EFL undergraduates participated in this study and all of them were Chinese native speakers in Hong Kong. The students responded to an online questionnaire designed to collect data on their use of and perceptions and beliefs about MT and its relationship between language learning. The results generally revealed that there was mix of views about MT, its capabilities, potential, quality, and viability. While some learners were skeptical about its accuracy and reliability, MT systems’ lack of efficiency in certain fields was not the reasons for not implementing MT tools. The respondents tended to rely more on electronic resources (e.g. free online tools and mobile apps) than non-electronic ones mainly because of its convenience, not accuracy. The findings showed positive perceptions and attitudes towards MT in language learning, in particular to promote students’ error detection and correction skills and to foster their comprehension skills and language awareness. It was found that free online machine translation (FOMT) tools (e.g. Google Translate) were the most accessible form of MT and commonly used by the majority of students. In line with prior studies (e.g. Jolley & Maimone, 2015; Nino, 2009), high percentages of respondents used MT on a regular basis for specific purposes, such as to verify hunches, to get gist of what a foreign text says and for help with vocabulary or terminology. They evaluated the overall accuracy of MT tools to be higher than its capacity to handle grammatical structures and thus it was reported that students infrequently relied on MT for dissemination purposes where output of a higher linguistic quality is required. Key findings included that genre or text type, translation segment length, and types of MT affected participants’ judgment of its quality. Despite a lack of training, students used MT tools critically. While a majority believed that MT could bring benefits to language learning, written assignments, and translation tasks, they were aware that it produced errors, especially when handling complex structures and long segments. Major strengths of MT pointed out by the respondents included widely available online, convenience and immediacy, good with lexical translation and simply-structured texts; however, literal translation, grammatical, syntactical and discursive inaccuracies, and unable to account for cultural references were regarded as its main weaknesses. Hence, students tended to consult other references such as online dictionaries, glossaries, and search engines for MT post-editing. The results also revealed a high level willingness and a keen interest among the learners to get familiar with and learn about MT. They expressed a desire for MT training, with a particular preference for system demonstration and hands-on practice of current MT software. It is suggested that it is importance for MT users to understand the potential and deficiencies of MT and raise their awareness as to the complexity of language learning and translation. This study shed light on pedagogical applications of MT in EFL learning. It also yielded insights and a preliminary framework for developing best practices and strategies for effective use of MT tools.