Numerous studies have suggested that neuropsychological test performance is affected by demographic variables such as age and education. This study examined the effects of age and education on the Trail Making Test, the Digit Symbol Sub-Test, and the Finger Tapping Test in a non-clinical sample of community dwellers with a relatively low level of education (8 to 12 years) in South Africa. The sample consisted of 161 participants across six age groups: 20-39, 40-59, 69-69, 70-79, 80-89 and 90-95 years. Results were examined for mean age effects and variability trends. Highly significant age effects were present across the age groups for all tests, however, there was no uniform pattern of variability across the tests. The Digit Symbol Substitution Test and the Finger Tapping Tests showed a pattern of increasing variability with increasing age, followed by a decrease in very old age while no trend was evident for the Digit Symbol extensions (the Immediate and Delayed Recall tests). The Trail Making Test, Parts A and B, showed a consistent trend of increasing variability across the age groups. Data from the present study was compared with existing data from two relatively high education samples, with equivalent age groupings, to examine education effects. Results showed an education effect for all tests with the high education groups outperforming the low education groups. Although the effects of education became less potent with advancing age, the mean performance of the oldest (80-89 years) high education age group was superior to that of the equivalent low education age group. Comparison of variability trends across both samples showed that the highest variability (the shuttle bulge) was present at the same point along the age axis, or at a later point, for the low education group, as that for the high education group. This finding is inconsistent with Jordan's (1997) 'shuttle model of variability' which predicts an earlier occurrence of the shuttle bulge (left shuttle shift effect) for a low education sample. This study demonstrated that performance on neuropsychological tests is influenced by age and education and highlighted the dangers inherent in unquestionably applying norms, which have not been corrected for age and education, when assessing the older adult.