There exist a group of children among cerebral palsied ones whose profile of the WISC or the WAIS exhibits a remarkable dip in scaled score in the block-design subtests. The dip seems to suggest their inadequacy in mental operations of spatial representations, as compared with the scaled scores of the other subtests. The present experiment was designed to investigate what kind of mental operations are required to perform the block-design test and to examine in what operations those cerebral palsied children have difficulty. An original and eight modified tasks of the block-design subtest in the WAIS were administered to four such cerebral palsied children. The Eight modified tasks were as follows: visual choice of a model among serevral similar designs(task 2); reproduction of a design from a visual model which was presented in a life-sized demonstration card(task 3); reproduction in which a model was presented in a life-sized demonstration card with divisional lines to clarify the block elements in the design(task 4); reproduc-tion of a design, by employing flat blocks(task 5); painting a copy of a model(task 6); painting a clopy of a model whos elemental divisions were presented one at a time(task 7); reproduction in which a model was constructed of real cubic blocks(task 8); and reproduction of a design after the experimenter demonstrated the same reproduction(task 9). The nine tasks involved six reproduction ones, and the six tasks were presupposed to be arranged orderly from more difficult to easier ones. It was found that the subjects often pass task 2 even when they fail in the original reproduction task, and that the subjects' reproduction improves in task 4 in most cases and somewhat improves in task 8 and task 5. These findings were considered in terms of the subjects' inadequacy in such internal operations as mental sectioning of a representation of a model and mental transformation between two and three dimensional representations.