This work presents a general sequential parameter-and-state estimation structure based on extents transformations for non-isothermal homogeneous reaction systems. The extents transformations allow for the development of a Linear Parameter-Varying (LPV) representation with a diagonal state matrix. The particular structure of the LVP system matrices and the physical interpretation of its parameters are used to propose an asymptotic estimator. The main advantage of this estimator is that its implementation does not depend on the chemical kinetic parameters. Based on the information given by the asymptotic estimators, two additional estimators are proposed: an adaptive estimator, and a Recursive Least Squares (RLS) estimator. The first one is used to estimate the chemical reaction rate, while the second to estimate the kinetic parameters. Convergence and tuning properties of the final structure are analyzed and tested on a CSTR example.