Humans reason by analogy (Lakoff and Johnson, 1980; Gentner, 1983; 2003; Hofstadter, 2001, 2006; Pinker, 2007). Some have further argued that analogs can be scientific models (Hesse, 1966, Clement, 1989) although clearly not all analogies are models. Analogies based on mere physical similarity are not equivalent to scientific models but analogies based on shared relationships between the analog and target maybe equivalent to scientific models. A literature review of analogies and scientific models indicates that all scientific models in geology are relational analogs. Relational analogs are equivalent to models because both: 1) are based on recognizing relational characteristics of the analog (or model) and target, 2) map similarities and differences between the analog (or model) and target and 3) support inferences about the target based on relational similarities between the target and analog (or model). Therefore, the cognitive processes involved in analogical thinking provide a theoretical, research-based framework for instruction designed to improve students' ability to learn how to use and generate scientific models.