This thesis is comprised of three studies concerning the use of an empirical version of the Esscher Transform for nonparametric option pricing. The first one introduces the empirical Esscher transform and compares its performance against some well-known parametric option pricing approaches. In our proposal, we make only mild assumptions on the pricing kernel and there is no need for a risk-neutral model. In the second study, we propose a method for nonparametric option pricing under a GARCH framework with nongaussian innovations. Several papers have extended nonparametric option pricing and provided evidence that this methodology performs adequately in the presence of realistic financial time series. To represent a realistic time series, we use a new class of observation driven model, called dynamic conditional score model, proposed by Harvey (2013), for modeling the volatility (and heavy tails) of the asset s price. Finally, in the third study, we introduce a new approach for indirect estimation of state-prices implicit in financial asset prices from empirical Esscher transform. First, we generalize the discrete version of the Breeden and Litzenberger (1978) method for the case where states are not equally spaced. Second, we use the historical distribution of the underlying asset s price and the observed option prices to estimate the implicit Esscher parameter.