<p>This paper investigates the trustworthiness of computation implemented in simulations in engineering, with a specific focus on equation-based simulations. Whereas opacity discussions in the computer simulation literature typically center on modeling opacity, we direct our attention to the grounds engineers should have for trusting that the computer indeed found an acceptable solution to a given model. This is a particularly sensitive issue in situations in which analytical or experimental methods are impossible or impractical. After critically reviewing alternative views found in the literature, we propose a systematic approach for evaluating the accuracy of solutions obtained via computational methods, that involves quantifying residual errors and evaluating the model’s conditioning. We argue that, in equation-based simulations, ensuring that the residual is small enough and that the model is sufficiently well-conditioned provides a sufficient epistemic foundation for engineers to place their trust in the results obtained via computer simulations as much as they trust the models they use, thereby giving a conceptual basis for this much needed kind of conditional trust.</p>