It is widely accepted today that, to learn English as a foreign language, leamersneed to amass a large amount of vocabulary. According to Lewis (1997), vocabularyis of vital importance to language acquisition and leaners ability to chunk languagesuccessfully is a key that helps them better understand how language works. However,it is discovered in Wanfang Database that, most of the research papers on lexicalchunks are focused on speculative or theoretical discussions on this issue (seeAppendix Ⅲ). Contrastively, actual effects oflexical chunks on English teaching andleaming and the disadvantages of misuse and overuse of lexical chunks are seldomexplored.　　 To compensate for these problems, the author carried out an empirical study onthe chunk leaming beliefs and strategies of Chinese non-English majorundergraduates, in an attempt to give a correlation analysis between subjects chunkcompetence and their College English Test Band-4 (hereinafter referred to as CET4)test performance. The subjects consisted of 102 non-English major sophomores whowere randomly selected on the premise that they had taken part in the CET 4 test inDecember, 2010. The procedures of the research consisted of a preliminary work, apilot study, a correlation analysis, a regression analysis and a summary of replies to anopen-ended question in the questionnaire.　　 This study led to four conclusions. Firstly, the Pearson correlation coefficientbetween students questionnaire scores and their CET4 scores stood at r=0.798. Andthe correlation was significant at the 0.01 level. Meanwhile, significance values ofsubjects questionnaire scores and CET4 scores were both p=.000, which indicatesthat there was a significant and positive correlation between subjects questionnairescores and CET4 scores. Secondly, it was also revealed by the results of thedescriptive analysis that that Chinese college students chunk competence was farfrom satisfactory. Most of them had problems such as insufficient chunk knowledgeor lack of chunk learning strategies, which hindered their efficiency and accuracy incommunication. Thirdly, from the results of the regression analysis, it was discoveredthat those subjects questionnaire scores and CET4 scores were linearly related. Thatis to say, we could possibly estimate subjects CET4 scores according to thecorresponding questionnaire scores . Fourthly, from subjects replies to theopen-ended question in the questionnaire, it was found that a considerable number ofstudents suggested that chunks should be learnt through use in real situations, whichindicated that indirect vocabulary learning activities (e.g. conversations, dialogues)should occupy more time than those of direct vocabulary learning (e.g. explicitvocabulary instructions).　　 This paper concludes by putting forward some pedagogical suggestions forlanguage teachers and implications for EFL learners in current English teachingsetting in China. It is expected that findings in this paper will be of certain referentialvalue for similar studies on the issue of lexical chunks.