This study examines the applicability of the modified Location Learning Test (mLLT) as a test of spatial memory in neuropsychological patients. Three groups of participants were examined: stroke patients, patients with diabetes mellitus and healthy participants (N=411). Three error measures were computed, the Total Score (index of overall performance), the Learning Index (the learning curve over subsequent trials) and the Delayed Recall Score, measuring decay over time. The Learning Index was the most sensitive measure, showing differences between the three groups as well as lateralization effects within the stroke group. Also, the mLLT correlated significantly with the Rey Auditory Verbal Learning Test, as well as with age and education level. Regression-based normative data were computed based on the healthy participants. In all, the mLLT appears to be a sensitive and valid test for the detection of object-location memory impairments in clinical groups.