Grid-based algorithms for quantitative association rule mining are high efficient, but they are fundamentally low dimension oriented. This paper extends the grid-based concept and proposes a metarule-guided generalized linked list-based algorithm aimed to mine multidimensional quantitative association rules from relational databases. Based on the metarule, the algorithm stores data tuples into the linked lists and mining is acted upon these linked lists. Experimental results show that our solution is size and dimensions scalable linearly. A math model is also introduced to endow the association rules with some prediction functions, which can be considered as an extension to the classification functions of association rules.