Testwise responses of grade-school students to multiple-choice science items and the intercorrelations between different testwiseness skills were examined in this document. A total of 134 third-grade, 133 fifth-grade, and 143 sixth-grade students from two elementary schools in a large suburban school district in Utah responded to 60 science items on tests developed specifically for each of these grades. Thirty of the items contained embedded clues intended to elicit one of five separate testwise behaviors. The other 30 science items were identical to the original items, but did not contain the embedded clues. Half of the students completed the items without embedded clues before the items with clues, while the rest of the students completed the items with clues first. The presence of embedded clues made no difference in the performance of third graders, but both fifth graders and sixth graders demonstrated an ability to use the embedded clues to improve their performance. Testwiseness skills related to verbal associations between the stem and correct response and to absurd response options were observed in fifth and sixth graders. The next skill to appear by the fifth grade was a skill related to recognizing synonymous or overlapping response options. None of the skills were intercorrelated among the fifth graders, but among sixth graders, the verbal associations skill and absurd options skill were moderately correlated. Testwiseness skills appear to develop independently of each other as students develop more complex cognitive processes. Testwiseness skills do not develop in all students at the same time. Sample test items are included, and 12 tables present study data. (Author/SLD)