This study developed a standardised mediated assessment to measure young children's mathematical ability in reasoning, abstraction and representation in number, computation, quantity, shape and relationship through six tasks with four levels. The percentage distribution of children at four levels on the tasks showed that the tiered mediations through varying the structural complexity, the form of presentation, the use of manipulatives and the explicitness of instructions were effective in reducing the difficulties of assessment tasks in differentiating and providing more information about young children's learning potential for learning mathematics. Furthermore, the results of the t-tests of independent samples indicated that mediated assessment could also provide this study with valid information on learning outcomes as static assessment. The implications and limitations of this study have been discussed. (Contains 5 tables and 3 figures.)