Intense cognitive-motor dual-tasks (CMDTs) typically induce central nervous system fatigue and reduce cognitive performance, even in individuals with high fitness levels. We aimed to explore the efficiency of CMDT training in preventing central fatigue and cognitive performance decline in such scenarios. Forty trained adults were randomly assigned to one of three high-intensity interval training (HIIT) groups (4 weeks, 2 sessions/week): cycling alone (HIITCTRL), cycling with cognitive training during low (HIITDTL), or high-intensity intervals (HIITDTH). Each training session consisted of 4 × 6 min self-selected intervals based on perceived effort (3 min "very strong", 3 min "very low"). Before and after training, participants completed an incremental cycling cognitive test (ICCT) until reaching task failure (TF). Sustained attention Mackworth score, knee-extensor neuromuscular peripheral (e.g., peak twitch) and central (voluntary activation, VA) indices, prefrontal cortex (PFC) oxygenation, cycling and mental perceived efforts were assessed at each stage. Data were interpolated at 20%-40%-60%-80% relative to TF pretraining (TFPRE). All groups increased ICCT endurance time after training (+3.0 ± 4.2 min, p < 0.001). After training, the Mackworth score increased at rest, 80% and TFPRE (+11% ± 2%, p < 0.001), alongside similar PFC oxygenation patterns (p = 0.13), whatever the group. HIITDTL and HIITDTH mitigated VA decline at TFPRE (+4.5% ± 3.6% and +5.0% ± 3.7%, respectively, p < 0.001) compared to HIITCTRL (+0.9% ± 4.0%). HIITDTL and HIITDTH reduced mental effort after training throughout the ICCT (-14 ± 11 and -19 ± 12, p ≤ 0.002) compared to HIITCTRL (-6 ± 13). Whereas a traditional HIIT program was associated with improvements in sustained attention alongside gains in endurance performance, CMDT-specific training more effectively mitigated central fatigue and reduced perceived mental effort during challenging CMDT scenarios.