The article addresses the need to devise assessment methods appropriate for task-based instruction in an Asian context. An initial attempt was made to use an existing (non-Chinese) system to establish the difficulty level of test tasks, and rating systems appropriate to them. This attempt, although promising, indicated a need for modifications to be viable in this new context. Consequently, a series of studies were undertaken, each building on its predecessor, which developed systems for assessing task difficulty and for rating the performances which resulted with test-tasks. A multi-category task-analysis scheme was devised, trialled and modified, and specific ratings scales were developed for each category. This scheme was used to assess task difficulty itself; and task performance, at a detailed subscale level, with these ratings being accomplished by ELT professionals and novices. The results indicate agreement, between task generators and task raters, on a three-step task difficulty scale. They also demonstrate satisfactory inter-rater agreement on performance itself, including agreement between ELT professionals and novices. The implications of this are discussed for the development of curriculum-linked task-based assessment, in general, and the usefulness of the scheme for class teachers focussed on developing within-class achievement tests.