This study describes heart rate (HR) responses during different small sided games (SSGs) in junior basketball players, and identifies the level of agreement between athlete and coach perceptions of internal training load calculated using the in-task rating of perceived exertion (RPE) method. Over a 6 week period, 12 male junior basketball players who played in the Spanish national under-18 League, played 7 games of one-a-side (1v1), 6 games of two-a-side (2v2), 8 games of five-a-side (5v5), and 5 games of superiority (3v2) situations. During 1v1, 2v2, 5v5, and 3v2 peak heart rates were 90.27 ± 3.37%, 92.68 ± 3.29%, 92.01 ± 3.48%, and 88.74 ± 5.77% of HRmax respectively. These differences were statistically significant between 1v1 and 2v2 (P<0.01), 1v1 and 5v5 (P<0.05), 2v2 and 3v2 (P<0.001), and 5v5 and 3v2 (P<0.001). Mean heart rate was 79.5 ± 4.4%, 83.1 ± 4.2%, 91.2 ± 4.7%, and 78.5 ± 7.5% of HRmax during 1v1, 2v2, 5v5, and 3v2, respectively, and differences were observed between 1v1 and 2v2 (P<0.001), 2v2 and 3v2 (P<0.001), and 5v5 and 3v2 (P<0.05). There were differences in athletes and coaches in-task RPE in all SSGs (all P<0.0001 apart from 5x5 P=0.0019). The 2v2 format elicited a higher mean in-task RPE in comparison to all other SSGs (P<0.001), possibly because 2v2 imposes a greater cognitive load.