This study explored the issue of evaluation in general and language testing in particular in the Algerian educational system. In precise terms, the present research investigated how a particular type of language tests, achievement tests, should be designed and developed in order to make them fit their intended purposes. This aimed to provide EFL teachers at the secondary school level in Biskra region with theoretical and practical understanding, with the expectations to make these developed tests become part of the learning experience and not a mere operation of grading for the sake of passing or failing. The rationale for this study is twofold. Firstly, given that there is a multitude of washback studies of language tests on teaching and learning, no empirically study of this sort was available for this particular context. Secondly, this study stems from the students’ low scores that do not actually reflect their performance by virtue of the non-pertinence between what these students learn and what they are tested in. Ultimately, the fundamental objective of the present exploration is to yield practical insights to overcome the array of anomalies present in the current testing system. In order to achieve this objective, an Alternative Testing Model was proposed. This model was first tried-out and then put into practice on a selected sample of EFL classes in Biskra region. In the investigation of the washback effects of the new testing system on participants’ perceptions of, and attitudes towards, teaching and learning, a mixed-methods approach was employed, using four data collection methods to gather insightful feedback. From the analysis of these data, the findings show that the new testing policy had positive washback effects on teachers and students in this study. It succeeded to change to some extent and with different degrees these participants’ attitudes and behaviours in teaching, learning, and assessment practices.