This study investigated the role of problem structure and metacognitive control in the analogical transfer of performance of 40 13- and 14-year-old gifted and highly gifted math students. Average and above average 16-, 17-, and 18-year-olds served as comparison groups. Students were given three algebra problems with solutions, followed by two target problems to be solved without a solution procedure. In Condition 1, none of the initial problems were related to either target, but in Condition 2, two of the initial problems were related to the target in deep structure or in surface structure. Metacognition was explored through analyses of videotaped think-aloud protocols and two standardized measures, the Learning and Study Strategies Inventory and the Action-Control Scale. No transfer effect was found for the highly gifted, but there was a significant effect for the gifted. Neither of the two high school groups showed positive transfer. Protocol analysis shows that high metacognitive control is associated with greater positive transfer, higher accuracy, and overall higher ability. Some negative transfer effects are discussed. Results from the standardized measures indicate that failure to recover from negative transfer may be related to effects of anxiety on metacognitive control (Contains 2 tables and 23 references.) (Author/SLD)