Findings from the fields of cognitive science and cognitive development propose a variety of evidence-based principles for improving learning. One such recommendation is that instead of having students practice solving long strings of problems on their own after a lesson, worked-out examples of problem solutions should be incorporated into practice sessions in Science, Technology, Engineering, and Mathematics (STEM) classrooms. Research in scientific laboratories and real-world classrooms has also identified a number of methods for utilizing worked examples in lessons, including fading the examples, prompting self-explanation of the examples, including incorrect examples, and providing opportunities for students to compare multiple examples. Each of these methods has been shown to lend itself well to particular types of learning goals. Implications for education policy are discussed, including rethinking the ways in which STEM textbooks are constructed, finding ways to support educators in recognizing and implementing effective cognitive science-based pedagogical techniques, and changing the climate in classrooms to include the perception of errors as a functional part of the learning process. [This article was published in "Policy Insights from the Behavioral and Brain Sciences," v2 n1 p24-32 Oct 2015.]