A major use of small-sided games (SSGs) in soccer training is the concomitant development of game-specific aerobic fitness. We hypothesize that the SSG formats of 2 vs. 2, 3 vs. 3, and 4 vs. 4 players reveal game-like intensities and therefore are most adequate to increase game-specific aerobic fitness. Heart rate (HR), percentage of maximum heart rate (HRmax), blood lactate concentration (La), and time-motion characteristics of 17 elite male youth soccer players (aged 14.9 ± 0.7 years, V[Combining Dot Above]O2max 61.4 ± 4.5 ml·kg·min, HRmax 199.6 ± 7.3 b·min) were collected by global positioning systems while performing the SSG formats. Repeated-measures analysis of variance and effect sizes were calculated to demonstrate the differences between SSG formats. Highest physiological responses were obtained in 2 vs. 2 (HR: 186 ± 7 b·min, HRmax: 93.3 ± 4.2%, La: 5.5 ± 2.4 mmol·L) followed by 3 vs. 3 (HR: 184 ± 8 b·min, HRmax: 91.5 ± 3.3%, La: 4.3 ± 1.7 mmol·L) and 4 vs. 4 (HR: 179 ± 7 b·min, HRmax 89.7 ± 3.4%, La: 4.4 ± 1.9 mmol·L). Pronounced differences were found for most physiological parameters and for time spent in the speed zones "walking" (<5.3 km·h), "moderate-speed running" (10.3-13.9 km·h), and "maximum sprinting" (≥26.8 km·h). The findings suggest that all the formats reveal game-like intensities and are suitable for aerobic fitness improvements. However, we found pronounced demands on the anaerobic energy supply in 2 vs. 2, whereas 3 vs. 3 and 4 vs. 4 remain predominantly on an aerobic level and differ mainly in the HR response. We suggest using 3 vs. 3 for soccer-specific aerobic fitness training.