This chapter introduces Computerized Adaptive Testing (CAT) with a particular focus on its application in STEM education for formative assessment. CAT leverages item response theory (IRT) models to deliver tailored test items that match individual students’ ability levels, thereby increasing efficiency, precision, and engagement. We first introduce the theoretical foundations of CAT, then discuss its architectural components, including item banks, initialization strategies, item selection algorithms, ability estimation methods, and stopping criteria. We further explore various CAT designs, such as item-by-item CAT, multistage testing, and hybrid CAT, as well as applications based on different psychometric models including unidimensional IRT, multidimensional IRT, and Cognitive Diagnosis Models (CDM). A case study illustrates the design and evaluation of a fixed-length CD-CAT for diagnosing mathematics skills in introductory physics courses, demonstrating that it can match or exceed the accuracy of traditional linear tests with fewer items. Finally, we highlight emerging research directions, including the use of generative AI for item generation, adaptive testing for complex skills, and open-source platform development to broaden CAT’s accessibility and impact in education.