Background: How exercise intensity affects working memory during concurrent motor-cognitive tasks remains unclear. This study examined working memory during dual-task cycling at the first and second ventilatory thresholds (VT1, VT2). Methods: Eleven trained male endurance athletes (VO_2max 49.38 ± 5.69 mL/kg/min) completed a graded exercise test and two randomized crossover sessions of cycling at VT1 or VT2 with a concurrent 2-back task, with 2-back assessments before and after exercise. Hits, false alarms, reaction times and signal detection indices (d′, c) were analyzed using repeated-measures ANOVA and Holm-corrected planned contrasts. Results: At both intensities, hits and d′ decreased during the dual-task (phase effects, p < 0.001), whereas the response criterion did not change. For hits, the primary outcome, the Condition × Phase interaction was significant [F(2, 20) = 4.90, p = 0.019, η²p = 0.329]: hits decreased substantially during the dual-task at VT2 (d = 1.21, pHolm = 0.010) but not significantly at VT1 (p = 0.074), and the VT1-VT2 difference during the dual-task was itself significant (d = 0.98, pHolm = 0.036). A planned contrast directly comparing the pre-to-dual-task change in hits between conditions supported this pattern (p = 0.022), although it did not survive the stricter Holm correction applied within that specific contrast family (pHolm = 0.089). Immediately after exercise, d′ remained below pre-task values after VT2 (pHolm = 0.008) but not after VT1. Conclusions: Concurrent cycling at VT2 was associated with a significantly larger acute decline in working memory than cycling at VT1, directly supporting the study’s primary hypothesis. This intensity-dependent pattern, observed in a modest sample, should be confirmed in larger, adequately powered studies.