There have been many comparisons of computational methods applied to ground states, but studies of organic reactions usually require calculations on transition states, and these provide a different test of the methods. We present calculations of the geometries of nineteen covalent-bond forming transition states using HF and twelve different functionals, including GGA, hybrid-GGA and hybrid meta-GGA approaches. For the calculation of the TS geometries, the results suggest that B3LYP is only slightly less accurate than newer, computationally more expensive methods, and is less sensitive to choice of integration grid. We conclude that the use of B3LYP and related functionals is still appropriate for many studies of organic reaction mechanisms.