Abstract: Grammar testing was discussed, including related concepts, its characteristics, application scenarios, and a comparison with traditional software testing. Test adequacy criterion of context-free grammar was formally defined and a family of grammar testing criteria was proposed. A formal analysis of the subsume relation between test adequacy criteria in the family was presented, and how the subsume relation between adequacy criteria guarantees a better fault detecting in the context of two classes of faults specific to grammar was also proved. Size complexity of the test adequacy criteria in the family was computed and compared. A grammar testing complexity metric was also developed, which can reflect the difficulty and effort to testing a grammar from the view of test adequacy.