A common account of working memory capacity, and its development, incorporates the notion of limited mental resources being shared between processing and storage components. Working memory tasks are assumed to measure this capacity. However, recent data from children's performance at one such task, counting span, implicate forgetting over time as being more important than the capacity for resource-sharing. Three experiments investigated the role of temporal factors in more depth and breadth by varying the retention requirements in working memory tasks while holding constant the overall processing difficulty. Counting span, operation span, and reading span tests in 6- to 11-year-olds were all shown to decline as the retention interval of stimulus items is increased and, across tasks, there were no consistent effects from different concurrent storage loads on processing speed. It is suggested that among children, performance in these tasks does not reflect a trade-off between resources allocated to processing and storage and that caution should be adopted in interpreting working memory span in adult populations.