In this paper, we develop a novel estimation procedure for parametric option pricing models specified under the risk-neutral measure. We set up our estimation strategy to minimize the distance in a functional sense between the option-implied and model-implied logarithm of conditional characteristic functions. Within a broad class of option pricing models, for which the characteristic function is marginally affine, the model’s latent state vector can be concentrated out in closed form. As a result, our estimation procedure is computationally fast and easy to implement, while at the same time exploiting all distributional information contained in an option panel about the risk-neutral dynamics of the underlying asset price. We establish the asymptotic properties of the parameter and state estimators and investigate the finite-sample performance of our method in Monte Carlo simulations. In an empirical application, we illustrate the usefulness of our estimation procedure based on by far the largest option panel, both in the time-series and cross-sectional dimensions, considered in the related literature.