Caffeine is commonly used to treat respiratory instabilities related to prematurity. However, the role of adenosinergic modulation and the potential long-term effects of neonatal caffeine treatment (NCT) on respiratory control are poorly understood. To address these shortcomings, we tested the following hypotheses: 1) adenosine A(1)- and A(2A)-receptor antagonists modulate respiratory activity at rest and during hypercapnia; 2) NCT has long-term consequences on adenosinergic modulation of respiratory control. Rat pups received by gavage either caffeine (15 mg/kg) or water (control) once a day from postnatal days 3 to 12. At day 20, rats received intraperitoneal injection with vehicle, DPCPX (A(1) antagonist, 4 mg/kg), or ZM-241385 (A(2A) antagonist, 1 mg/kg) before plethysmographic measurements of resting ventilation, hypercapnic ventilatory response (5% CO(2)), and occurrence of apneas in freely behaving rats. In controls, data show that A(2A), but not A(1), antagonist decreased resting ventilation by 31% (P = 0.003). A(1) antagonist increased the hypercapnic response by 60% (P < 0.001), whereas A(2A) antagonist increased the hypercapnic response by 42% (P = 0.033). In NCT rats, A(1) antagonist increased resting ventilation by 27% (P = 0.02), but the increase of the hypercapnic response was blunted compared with controls. A(1) antagonist enhanced the occurrence of spontaneous apneas in NCT rats only (P = 0.005). Finally, A(2A) antagonist injected in NCT rats had no effect on ventilation. These data show that hypercapnia activates adenosinergic pathways, which attenuate responsiveness (and/or sensitivity) to CO(2) via A(1) receptors. NCT elicits developmental plasticity of adenosinergic modulation, since neonatal caffeine persistently decreases ventilatory sensitivity to adenosine blockers.