For decades traditional methods of testing have been criticized for saying relatively little reliably about students' ability as well as causing anxiety, which can negatively affect students' recall of learned information. The reform movement with its innovative approaches focusing on learner-centered education perceives assessment as an interactive feedback mechanism, which must provide for active, collaborative reflection by both teacher and students. This means that students must be active participants in designing assessment tasks and be given responsibility for using assessment data to monitor and improve their own learning (Valencia, 1990, p. 339). Focusing on alternative methods of assessment proposed by the opponents of traditional ones, the present study aimed at investigating the impact of cooperative test construction on Iranian EFL (English as a Foreign Language) students' achievement as well as their attitudes towards such tests. The participants in this study were second- grade high school students (equivalent to Grade 11 in US senior high school) who were assigned to experimental and control groups based on their scores on a standardized retired version of Nelson test. Both groups received the same schedule of instruction for sixteen weeks. The students in the experimental group experienced cooperative test construction while the students in the control group did not have any role in the construction of their tests. The findings revealed statistically significant difference between grammatical knowledge of the students who cooperatively made their own test items and that of those who were tested traditionally. Furthermore, the students' reactions to cooperative test construction were overwhelmingly positive.