This study researched the nature of tasks used in Grade 10 mathematics tests and end-of-year national examinations. The study was carried out in three, purposively selected, Grade 10 schools in the Ohangwena region in Northern Namibia. For the purpose of this study, a mixed method approach was employed to analyse a combination of both quantitative and qualitative data. A sample of three tests per mathematics teacher from the three participating schools and national examinations question papers for the past three consecutive years (2011-2013) were analysed using Stein, Smith, Henningsen, & Silver’s (2000) framework of cognitive demand. The study was divided into two phases. Phase 1 was the analysis of teacher test items and national examination items in terms of their cognitive demand. Phase 2 involved semi-structured interviews with three selected teachers to probe their views and find out their basis for selecting test items. The findings of this study revealed that there was no substantial difference in the distribution of the levels of cognitive demand in both tests and national examinations items. The study, however, showed that mainly tasks requiring only procedures without connections dominated the tests and the examinations. The number of higher level tasks in both tests and examinations analysed was low. There was no single task coded at level 4 in any of the teachers’ tests. Only 2% of tasks could be classified at level 4 in the examination items. The study also revealed that since tests and examinations assess the same learning objectives from the syllabus, most of the test items set by teachers were extracted from the national examinations question papers. The paper recommends that more tasks at a higher level category need to be included in assessment tasks to promote critical thinking amongst learners.