The aim of this study was to examine to what degree fluid intelligence can be modified in aged subjects. The effectiveness of a cognitive training program designed to enhance one primary component of fluid intelligence, Figural Relations, was assessed by comparing the posttraining performances of 15 experimental and 15 control subjects (mean age: 69; age range 59-85) using a transfer paradigm and three posttraining assessments conducted approximately 1 week, 1 mo, and 6 mo following training. The post-training performance of the two groups was compared on three near (fluid intelligence) and one far (crystallized intelligence) transfer measures. A hierarchical pattern was predicted with the magnitude of training effects ordering themselves in descending order from near to far transfer measures. The training program was successful in enhancing performance on the fluid-nearest measure on all three posttests and for the next fluid-near measure on the first posttest. In addition, significant retest effects resulted which, however, were neither task-specific nor hierarchically ordered, but general and therefore indicative of ability-extraneous factors, such as test sophistication. The findings contribute to a position implying that intellectual performance in old age is more modifiable through short-term behavioral intervention than traditionally assumed.