In parameter estimation for systems described by ordinary differential equations, the region of convergence for the Gauss-Newton method can be substantially enlarged through the use of an "Information Index" and an optimal step-size policy. The information index provides a measure of the available sensitivity information as a function of time, thereby locating the most appropriate section of data to be used for parameter estimation. The use of the chosen section of data significantly improves the conditioning of the linear algebraic equations that are solved at the end of each iteration In the Gauss-Newton method, and the region of convergence is thus expanded. If observations are unavailable in the most appropriate section of data, artificial data can be generated by data smoothing and interpolation. The effectiveness of the approach is illustrated by four examples depicting typical chemical engineering systems. The proposed procedure is especially attractive for systems described by stiff differential equations.