Since the New Curriculum Standards clearly encouraged the use of the TBLT in the Chinese mainland in 2001, it has attracted more and more attention from scholars, researchers and front-line teachers. Yet, many of these researches have been conducted under laboratory conditions or in tightly controlled settings. The study aims to find out what the real role of the TBLT in ordinary classrooms, how front-line teachers make use of it, and how they balance tasks and language focus. Hereby, the author observed three primary teachers‘ classes in Xiangzhou Experimental School for three weeks, and collected data. First, the traditional Grammar Translation Method (GT), Direct Method (DM), Audio-lingual Method (ALM), Total Physical Response (TPR), Communicative Language Teaching Method (CLT), etc., have been reviewed. Then, the definition of ―task‖, varieties of tasks, the differences between tasks and exercises are carefully studied. Next, its theoretical support is analyzed from two perspectives: second language acquisition and social-constructivism. Last, the author concludes by identifying five features of the TBLT: authenticity, communication, language dependency and task completeness, ―learn by doing‖, and cooperation. Methodologically, the study has adapted the Class Observation Scheme for Communicative Orientation of Language Teaching for data analysis, (Allen, Frölich and Spada, 1985; Chaudron, 1988; cited in Nunan, 1996) and has found that one of the teachers combined TPR with TBLT, the second teacher used CLT and TBLT, and the third teacher didn‘t use TBLT at all. Teachers‘ confusion about tasks and exercises, teacher-centred class, and students‘ reluctance or inability to use English caused difficulties in task implementation. The research also found: 1) students were interested in tasks, partly because they were motivated by bonus marks, rather than completely task-driven; 2) fun is primary for primary students rather than ―meaning is primary‖, so teachers can do something entertaining in class in order to make them interested in tasks; 3) pair/group work can, to some extent, solve the problems caused by the large class size; 4) teachers must direct students‘ attention to language focus, otherwise, students will not focus on language automatically when they are doing tasks; 5) teacher might as well provide inputs that are in advance a little of students‘ current level, and give a push iii when students produce language, and then students can better acquire language; 6) without authentic context, real communication are not likely to happen. Thus, the study suggests: 1) teachers should try the strong form of TBLT and increase the difficulty level of tasks occasionally; 2) students should be more independent from teachers and their mother tongue, and responsible for their learning; 3) authentic context, good rapport between teachers and students, good classroom management, rich input, a push for high-quality output, and appropriate entertaining activities are necessary for a successful TBLT class. Key words: TBLT; tasks; language focus