Retest reliability and validity of a computational method (VS, [19]) for the analysis of visual spatial perception in brain damaged patients was investigated. Retest reliability was tested in 20 patients (retest interval: 1 week; range: 1-3) and varied from r = 0.65-0.95 for the different subtests for the "point of subjective equality" and r = 0.39-0.89 for the "interval of uncertainty". In the several investigations concerning the validity of VS. 234 patients and 10 normal subjects participated. In the first place, we found significant relationships of some of the subtests of VS to similar visual-spatial tests, but no relationship to tests of visual object perception. Secondly, we found significant relationships to certain tasks which approximated daily functioning. In summary, the present results demonstrate the objectivity, retest-reliability and validity of VS in brain damaged patients.