Scaling, to remove the effects of body size, is an important methodological approach for enabling an equitable comparison of performance differences between individuals who vary in anthropometric characteristics. Many previous studies using scaling in sport have done so based on only one or two anthropometric characteristics, with only one study to date adopting a three-dimensional approach. To apply a three-dimensional allometric model to rowing ergometer performance (REP) in adolescents, and to detect whether key 'scaling' parameters remain stable when scaling REP both before and after a 6-week training intervention. Novel three-dimensional allometric models were used, incorporating body mass, stature and waist circumference (WC) to detect the most appropriate body size dimension(s) and scaling parameters associated with REP before and after a 6-week training intervention. Using this more flexible and sensitive three-dimensional allometry demonstrated that, following 6-weeks of training, there was a change in the ideal body shape associated with REP. Before training, taller, but not heavier, adolescents performed better. After 6-weeks of training, older participants with a greater body mass but smaller WC performed better. Scaling approaches are important for evaluating performance differences between individuals of differing body size. The findings from the current study (using a novel three-dimensional allometry approach) emphasise that relatively subtle changes in individuals' behavioural characteristics, such as changes in their training/fitness status, can result in quite dramatic changes in the body dimension characteristics and scaling parameters deemed to be key for performance in activities such as REP.