Abstract: A test consisting of 7 word problems that were problematic from a realistic point of view was administered to 207 pupils in Grade 6.The subjects were first asked to solve the problems themselves,and afterwards to evaluate four different answers given to each word problem.The results showed that:(1)The percentage of pupils having realistic considerations is 36.6%,higher than the percentage of pupils having realistic answers(27.3%)significantly;(2)In realistic considerations,the high-achievers were better than average-achievers significantly,the average-achievers were better than low-achievers significantly.In evaluating realistic answers "impossible to know",the high-achievers were higher than average-achievers and low-achievers significantly,while in evaluating no-realistic answers they were lower than average-achievers and low-achievers significantly;(3)A significant increase of realistic considerations was due to the instructional condition.And there was significant interaction between instructional factor and mathematical performance.As in evaluating the realistic answers "impossible to know" and no-realistic answers,the instructional condition had no influence on the subjects.

Key words: realistic considerations, problematic word problems, instructional condition