The paper proposes two different estimation procedures for nonlinear panel data models with a general parametric heterogeneity distribution. Using small—sigma and the Laplace approximation, easily computable analytical solutions to the marginal likelihood are presented.