Sleep is vital, but its benefits are unknown. One hypothesis posits that sleep is an antioxidant defence mechanism. Indeed, reactive oxygen species (ROS) have been shown to accumulate during prolonged waking, and experimentally increasing the ROS load of the brain induces sleep. In Drosophila melanogaster, a crucial ROS sensor in sleep-promoting dorsal fan-shaped body neurons (dFBNs) is the β-subunit, termed Hyperkinetic (Hk), of the voltage-gated potassium channel Shaker. Oxidation of a stably bound NADPH cofactor causes conformational changes in the channel, which increase dFBN activity. dFBNs could thus act as an early-alert system for rising ROS levels in the brain. We investigated sleep’s possible role as an antioxidant defence by either curbing the production of ROS or altering the ROS-sensing abilities of dFBNs. To achieve this, we analysed sleep patterns and common symptoms of chronic sleep deprivation, memory impairment and early death, of Hk mutants and flies expressing the ROS-limiting alternative oxidase (AOX) in their mitochondria. Both manipulations caused significant sleep losses, but only short-sleeping Hk mutants showed impaired learning and shortened lifespan, while both were unaffected in short-sleeping flies expressing AOX pan-neuronally. Rescue of Hk expression in dFBNs was sufficient to restore normal sleep and learning, but not lifespan, indicating that lifespan is dependent on Hk expression outside of dFBNs. Similarly, AOX expression in dFBNs only rescued learning, but not lifespan. Our experiments show that proper ROS sensing is essential for regular sleep duration, while lowering brain-wide ROS production reduces sleep without producing symptoms of chronic sleep deprivation.