Sequential testing with brief cognitive tools has been recommended to improve cognitive screening and monitoring, however the few available tools still depend on an external evaluator and periodic visits. We developed a self-administered computerized test intended for longitudinal cognitive testing (Brain on Track). The test can be performed from a home computer and is composed of several subtests, expected to evaluate different cognitive domains, all including random elements to minimize learning effects. An initial (A) and a refined version of the test (B) were applied to patients with mild cognitive impairment or early dementia (n = 88) and age and education-matched controls. A subsample of a population-based cohort (n = 113) performed the test at home every three months to evaluate test-retest reliability. The test's final version Cronbach's alpha was 0.90, test scores were significantly different between patients and controls (p = 0.001), the area under the receiver operating characteristic curve was 0.75 and the smallest real difference (43.04) was lower than the clinical relevant difference (56.82). In the test-retest reliability analysis 9/10 subtests showed two-way mixed single intraclass consistency correlation coefficient >0.70. These results imply good internal consistency, discriminative ability and reliability when performed at home, encouraging further longitudinal clinical and population-based studies.