This study investigated the potential of dynamic testing of geometric analogical reasoning in differentiating between the potential for learning of gifted and average-ability children (aged 9-10 years old). In doing so, it was analysed whether planning, a higher-order executive function, was related to outcomes of the dynamic test, and to instructional needs during training. Employing a pretest-training-post-test control group design, participants were split into four subgroups: gifted dynamic testing (n = 24), gifted control (n = 26), average-ability dynamic testing (n = 48) and average-ability control (n = 50). The results revealed that children who were dynamically tested progressed more in accuracy from pre-test to post-test than their peers who received practice opportunities only. Gifted children outperformed their average-ability peers in accuracy, but showed similar levels of improvement after training or practice only. Moreover, gifted children showed they needed fewer prompts during training than their average-ability peers. Planning was found to be related only to pre-test accuracy, and the number of prompts needed at the first training session, but not to post-test accuracy or the number of prompts needed at the second training session. In the discussion, educational implications of the findings were discussed.