Scientific literacy and authenticity have gained a lot of attention in the past few decades worldwide. The goal of the study was to develop various authentic assessments to investigate students' scientific literacy for corresponding to the new curriculum reform of Taiwan in 1997. In the process, whether ninth graders were able to apply school knowledge in real-life problems was also investigated. Over the course of our two-year study, we developed authentic assessments to investigate a stratified random sampling of 1,503 ninth graders' levels of scientific literacy, including scientific cognition, process skills, application of science, habits of mind, nature of science, and attitude towards science. The purpose of this article is to discuss three different formats of authentic assessments: multiple-choice, open-ended, and hands-on test items, which we developed to investigate scientific cognition. To validate the three formats of authentic assessments, students' performance on these three assessments were compared with the science section of Taiwan's Academic Attainment Testing (STAAT), and the values of Pearson correlation coefficient were all at the significant level, ranging from 0.205 to 0.660 (p less than 0.01). We found that our three authentic assessments were better in evaluating students' authentic abilities in science than standardized tests (such as STAAT). Further authentic assessments, particularly the hands-on activity, benefited low-achieving students. Concerning the common themes tested in the authentic assessments, students performed better in a multiple-choice test than an open-ended test on electricity and heat and temperature. In addition, two themes of chemical reactions and reactions of acid and base with indicators were performed best in a hands-on test than in the other two tests. In this article, we provide evidence that authentic assessments could be developed in different formats to investigate students' scientific cognition as part of the national test. Of these formats, the multiple-choice, open-ended, and hands-on test items are all shown to be sensitive in their evaluation of students' cognition in science. (Contains 5 tables and 4 figures.)