Focuses on the second order asymptotic results for likelihood-based inference in Gaussian non-linear regression models. Density of the maximum likelihood estimator; Aspects of approximate ancillary statistics; Derivation of statistics to test the hypothesis on one component of the parameter.