Model-based revision of knowledge bases ex pressed as first-order theories was shown in [Winslett 88b] to be useful as a means of de scribing and reasoning about the effects of ac tions. This paper shows that model-based the ory revision is actually expressible as a form of circumscription. This shows that in cer tain applications, the cumbersome conceptu ally machinery of circumscription can be re placed by the intuitively simpler ideas of modelbased theory revision. Where theory revision techniques are insufficient to capture the se mantics of change in an application, circum scription will offer a more flexible environment. In addition, future advances in computing cir cumscription can be mapped to improvements in computing theory revisions, and vice versa.