The assessment of mental effort is increasingly relevant in neuro-cognitive and lifespan domains. Pupillometry - the measure of the pupil size - is often used to assess effort but has disadvantages. Analysis of eye movements may provide an alternative, but research has been limited to easy and difficult task demands in younger adults. An effort measure must be sensitive to the whole effort profile, including 'giving-up' effort investment, and capture effort in different age groups. The current study comprised three experiments in which younger and older adults (both sexes) listened to speech masked by background babble at different signal-to-noise ratios associated with easy, difficult, and impossible speech comprehension. We expected individuals to invest little effort for easy and impossible speech ('giving up') but to exert effort for difficult speech. Indeed, pupil size was largest for difficult, but lower for the easy and impossible speech conditions. In contrast, gaze dispersion decreased with increasing speech masking in both age groups. Critically, gaze dispersion during difficult speech returned to levels similar to easy speech after sentence offset, where acoustic stimulation was similar across conditions, whereas gaze dispersion during impossible speech continued to be reduced. These findings show that a reduction in eye movements is not a byproduct of acoustic factors, but instead suggest that neuro-cognitive processes, different from arousal-related systems regulating the pupil size, drive reduced eye movements during high task demands. The current data thus show that effort in one sensory domain (audition) differentially impacts distinct functional properties in another sensory domain (vision).