Solving test items might require abilities in test-takers other than the construct the test was designed to assess. Item and student characteristics such as item format or reading comprehension can impact the test result. This experiment is based on cognitive theories of text and picture comprehension. It examines whether integration aids, which relate pictorial representations to the corresponding textual representations in item stimuli, affect performance in a science test. The results show that items containing referential connections between both representations and highlighting associated information are easier to solve than non-integrated items (i.e., items without aids). However, this is only true for information-complementary representations, not for information-equivalent representations. Furthermore, an effect of reading comprehension on students' test performance observed when complementary information was presented in a non-integrated format was absent in the integrated format condition.