The chief hindrance to the widespread adoption of attribute-grammar-based systems has been that they are profligate consumers of storage. This paper concerns new storage management techniques that reduce the amount of storage used by reducing the number of attribute values retained at any stage of attribute evaluation; it presents one algorithm for evaluating an $n$-attribute tree that never retains more than $O(\sqrt{n})$ attribute values, and it presents a second algorithm that never retains more than $O(\log n)$ attribute values.