Casamichana, D, Castellano, J, Díaz, AG, and Martín-García, A. Looking for complementary intensity variables in different training games in football. J Strength Cond Res XX(X): 000-000, 2018-The main aim of this study was to identify which combination of external intensity training load (iTL) metrics capture similar or unique information for different training game (TG) formats and official ... [Show full abstract] matches (OMs) in football using principal component (PC) analysis. Ten metrics of iTL were collected from 24 professional male football players using global positioning technology. A total of 348, 383, 120, 127, 148, and 207 individual files for small-sided possession games, medium-sided possession games, small-sided games, medium-sided games, large-sided games, and OMs, respectively, were studied. Principal component analysis was conducted on each game format. Extraction criteria were set at an eigenvalue of greater than one. Varimax rotation mode was used to extract more than one PC. Intensity training load metrics with PC "loadings" above 0.7 were deemed to possess well-defined relationships with the extracted PC. In each TG and OM, 3 PCs were identified. For the first PC, eigenvalues for each game format ranged from 3.89 to 4.45, which explained 39-44% of the information (i.e., variance) provided by the 10 iTL metrics. For the second PC, eigenvalues ranged from 2.17 to 2.47, explaining 22-26% of iTL information. For the third PC, eigenvalues ranged from 1.41 to 1.98, explaining 14-20% of iTL information. This would suggest that TG and OM have multidimensional demands; so, the use of only a single iTL could potentially lead to an underestimation of the physical demands. Consequently, a combination of 3 iTL metrics is required during professional football game formats.