Using the MP2 method and basis sets up to near Hartree-Fock limit (NHFL) quality selected to saturate the s, p, d and f shells, we have evaluated the validity of the counterpoise method (CP) as a way of correcting the basis set superposition errors (BSSEs) for hydrogen bonded systems for some of the most popular basis sets, extending previous studies to larger basis sets. As test systems we have chosen two geometries of the hydrogen fluoride dimer, the minimum and a point in the repulsive region. It is shown that the CP method leads to an accurate description of the interaction energies, as long as the basis set employed has the minimum quality required for a description of the physical forces under consideration. Otherwise, it tends to overcorrect the BSSE. In general, the CP method gives interaction energies which depend less on the selected basis set. Finally, a basis set leading to a small BSSE does not necessarily provide a good description for strongly interacting systems, and the BSSEs of the NHFL basis sets employed here are still not negligible.