A parametric model of stochastic production is proposed and demonstrated. The parametric model treats output as a random variable with a distribution that is conditional on inputs. Maximum likelihood estimation of the model is shown to produce consistent and asymptotically efficient estimates.