This study compared 19 older adults and 20 younger adults on the Groton Maze Learning Test((c)) (GMLT), a novel computerized hidden maze learning test that assesses processing speed, spatial learning efficiency, and error monitoring. Convergent validity of this test was assessed by comparing GMLT scores to Paced Auditory Serial Addition Test (PASAT) and Tower of Toronto (TOT) scores. In the full sample, all GMLT measures correlated strongly with both PASAT and TOT scores (r's=0.53 to 0.73). GMLT measures most sensitive to detecting between-group differences were the Timed Chase Test (TCT), legal errors, and perseverative errors (Cohen's d's=3.81, 2.40, and 2.40, respectively). Scores on the visuomotor processing speed subtest of the GMLT attenuated the relationship between age group and maze efficiency index scores, but not perseverative and "rule-break" errors. These results suggest that normal aging is associated with impaired performance on a novel computerized measure of spatial learning efficiency and error monitoring, and that processing speed attenuates the relationship between age and spatial learning efficiency.