Analogical reasoning is assumed to play a large role in learning and problem solving in everyday and school settings. It was examined whether a newly developed dynamic test of analogical reasoning would be sufficiently difficult for identifying young gifted children's potential for solving analogies. The study included 74 gifted (n = 31) and average-ability (n = 43) children between 7 and 8 years old. Employing a pre-test--training--post-test format, in which half of the children received a graduated prompts training and the other half received a control task between pre-test and post-test, it was investigated (1) whether trained children would improve more in accuracy from pre-test to post-test than their untrained peers, and whether (2) gifted and average-ability children would demonstrate differences in their level of improvement from pre-test to post-test, and (3) their needs for instruction during training. The results indicated that dynamically tested children improved more than those in the control condition. In addition, the dynamic test seemed sufficiently difficult for the gifted children: regardless of whether they were trained, gifted children demonstrated superior accuracy scores than their average-ability agemates at pre-test and post-test, but similar levels of improvement. They were also found to need fewer instructions during training.