We present an investigation of basis set and correlation effects on transition stale geometries and primary and secondary kinetic isotope effects within an E2 model system. Four different ab initio methods have been employed : Hartree-Fock (HF), second- and partial-fourth-order Moller-Plesset perturbation theory, and quadratic configuration interaction with single and double excitations. Calculations have also been performed using density functional theory methods, but they perform poorly on this system. Eleven different basis sets up to 6-311++G(2df,2p) have been employed at the HF and MP2 levels. Both geometries and kinetic isotope effects are influenced by the theoretical level, and the HF method gives results significantly different from the correlated methods. More important, the HF method produces wrong relative values of both the primary and one of the secondary kinetic isotope effects. A uniform scaling of the HF frequencies is shown to have varying effects, improving the performance for PKIEs significantly, but sometimes increasing the disagreement for SKIEs with the best theoretical method. To obtain results comparable to the largest calculations done, it is necessary to include electron correlation and employ a basis set of at least 6-31+G(d) quality.