In a computing environment with multiple work queues and limited resources, an initial assignment of computing threads is given to each work queue. The utilization of these initially allocated computing threads is monitored as they perform computing jobs, and work queues with unused computing threads assigned to them may have those unused computing threads reassigned to a shared pool, where other work queues which are fully utilizing their assigned computing threads can temporarily borrow computing threads to complete additional computing jobs assigned to them. Work queues cannot borrow from the thread pool until they have recalled all of their reassigned computing threads back from the thread pool.