The goal of this study was to (i) assess the physical and anthropometric differences between three levels of climbers and (ii) predict climbing ability by using a multiple regression model. The participants were divided into novice (n = 15), skilled (n = 16), and elite (n = 10) climbers. Anthropometric characteristics such as height, weight, percentage of body fat and muscle, bi‐acromial breath, arm span, and ape index were measured. General and specific strength were assessed through an arm jump test, a bench press test, and a hand and finger grip strength test in maximal and endurance conditions. All variables were combined into components via a principal component analysis (PCA) and the components used in a multiple regression analysis. The major finding of this study is that climbing ability is more related to specific rather than general strength. Only finger grip strength shows a higher level of initial strength between all samples while the arm jump test discriminates between climbers and non‐climbers. The PCA reveals three components, labeled as training, muscle, and anthropometry, which together explain 64.22% of the variance. The regression model indicates that trainable variables explained 46% of the total variance in climbing ability, whereas anthropometry and muscle characteristics explain fewer than 4%.

The goal of the present study was to validate a new ecological power-test on athletes of different levels and to assess rock climbers' profiles (boulderers vs. route climbers). 34 athletes divided into novice, skilled and elite groups performed the arm-jump board test (AJ). Power, time, velocity, and efficiency index were recorded. Validity was assessed by comparing the distance with the value extracted from the accelerometer (500 Hz) and the reliability of intra- and inter-session scores. Moreover, a principal component analysis (PCA) was used to assess the climbers' profiles. The AJ test was quite valid, showing a low systematic bias of -0.88 cm (-1.25%) and low limits of agreement (< 6%), and reliable ( Intra-class correlation coefficient = 0.98 and CV < 5%), and was able to distinguish between the 3 samples (p < 0.0001). There was a good correlation between relative upper-limb power (r = 0.70; p < 0.01) and the AJ score. Moreover, the PCA revealed an explosive profile for boulderers and either a weak and quick or slow profile for route climbers, revealing a biomechanical signature of the sub-discipline. The AJ test provides excellent absolute and relative reliabilities for climbing, and can effectively distinguish between climbing athletes of different competitive levels. Thus, the AJ may be suitable for field assessment of upper limb strength in climbing practitioners.