Matrix isolation studies in solid argon and neon at 4.2 K reveal that iodoacetic acid initially only exists as its ground state (c,x) conformer with an almost perpendicular I-C-C=O dihedral angle, but UV irradiation in the 240-255 nm range leads to population of the 0.8 kcal mol(-1) less stable (c,c) isomer. The latter structure exhibits a close 3.23 angstrom contact of the iodine and carbonyl oxygen atoms decidedly below the sum of their van der Waals radii (3.50 angstrom). Increasing the matrix temperature by only a few Kelvin triggers the thermal back reaction of (c,c) to (c,x) and leads to an estimated upper limit of 0.38 kcal mol(-1) for the associated torsional barrier. While wave function methods including completely uncorrelated Hartree-Fock theory have no problem to identify (c,c) as a proper minimum, many popular density functionals fail to describe the C-C torsional potential in cis-iodoacetic acid qualitatively correct. We assessed the performance of 12 density functionals of different levels of sophistication, namely, the BLYP, PBE, TPSS, B3LYP, BHandHLYP, PBE0, M06-2X, CAM-B3LYP, omega B97X-D3, B2-PLYP, B2GP-PLYP, and DSD-PBEP86 methods, against accurate extrapolated CCSD(T)/CBS(T-Q)//MP2/def2-TZVPP energies and found that almost all of them yield acceptable relative energies. Still, even some of the best performers fail to find a reasonably deep minimum in the region of the (c,c) conformer, and addition of the empirical D3-dispersion correction does not remedy the qualitative shortcoming. Instead, inclusion of a sufficient amount of (long-range) exact exchange and likely a proper treatment of medium-range correlation effects all along the torsional coordinate play an important role in the proper description of the sub-van der Waals iodine-oxygen contact. More modern, recommended functionals do not suffer from the described shortcoming.