When interacting with a moving object, observers typically keep their eyes on the object to continuously update its position and integrate motion information. Yet, in real-world settings, such as in traffic, observers may be required to simultaneously interact with multiple moving objects. Here, eye-hand coordination was tested in human observers that used a Kinarm robot to intercept or avoid multiple falling objects. In the ‘hit’ condition, observers attempted to intercept all objects, while the number of objects that simultaneously moved down the screen gradually increased from 1 to 8. In the ‘hit and avoid’ condition, observers attempted to hit filled targets and avoid open targets. The ratio of ‘hit’ to ‘avoid targets’ was always 2-1 and the number of falling objects increased from 3 to 6 to 9. In the ‘hit condition’, interceptive performance decreased when more than 4 objects had to be intercepted simultaneously. Observers attempted to track each object with foveal (i.e., high acuity) vision and were better at intercepting foveated objects compared to objects that were tracked peripherally. In the ‘hit and avoid’ condition, observers consistently intercepted ‘hit targets’, while eluding ‘avoid targets’. Moreover, observers used peripheral vision to recognize ‘avoid targets’ whenever possible, making foveal vision available for tracking the ‘hit targets’. These results indicate that observers attempted to track each moving object with foveal vision to aid accurate object interception. Critically, patterns of eye hand coordination—i.e., whether and how long each object was tracked—flexibly adapted with changing visuomotor task demands.