A study in Sri Lanka concerning the effects of second language tests, specifically the O-Level examination in English as a Second Language, on classroom language instruction is reported. The study investigates the phenomenon of washback or backwash, the influence of testing on instruction. It is cited as the only known research investigating washback in language education through classroom observation. The study was conducted at the secondary school level, and combined classroom observation with data from interviews, questionnaire responses, and test analyses to determine whether washback exists, to what degree it operates, and whether it is a positive or negative force in this educational context. The report gives background information on the project; discusses the characteristics of positive and negative washback in terms of instructional content, instructional methods, and techniques, and assessment and presents the results of two rounds of classroom observation. It is concluded that washback occurred in both positive and negative forms, to some degree, in teaching content, but not in methodology. Evidence of washback, both positive and negative, on the way teachers and local education officers design tests was also found. An 11-item bibliography is appended. Further research is recommended. (MSE)