Inferences in measurement error models can be sensitive to modeling assumptions. Specifically, if the model is incorrect then the estimates can be inconsistent. To reduce sensitivity to modeling assumptions and yet still retain the efficiency of parametric inference we propose to use flexible parametric models which can accommodate departures from standard parametric models. We use mixtures of normals for this purpose. We study two cases in detail: a linear errors-in-variables model and a change-point Berkson model.