Intramolecular H-atom transfer in model peptide-type radicals was investigated with high-level quantum-chemistry calculations. Examination of 1,2-, 1,3-, 1,5-, and 1,6[C ↔ N]-H shifts, 1,4- and 1,7[C ↔ C]-H shifts, and 1,4[N ↔ N]-H shifts (Scheme 1), was carried out with a number of theoretical methods. In the first place, the performance of UB3-LYP (with the 6-31G(d), 6-31G(2df,p), and 6-311+G(d,p) basis sets) and UMP2 (with the 6-31G(rf) basis set) was assessed for the determination of radical geometries. We found that there is only a small basis-set dependence for the UB3-LYP structures, and geometries optimized with UB3-LYP/6-31G(d) are generally sufficient for use in conjunction with high-level composite methods in the determination of improved H-transfer thermochemistry. Methods assessed in this regard include the high-level composite methods, G3(MP2)-RAD, CBS-QB3, and G3//B3-LYP, as well as the density-functional methods B3-LYP, MPWB1K, and BMK in association with the 6-31+G(d,p) and 6-311++G(3df,3pd) basis sets. The high-level methods give results that are close to one another, while the recently developed functionals MPWB1K and BMK provide cost-effective alternatives. For the systems considered, the transformation of an N-centered radical to a C-centered radical is always exothermic (by 25 kJ · mol -1 or more), and this can lead to quite modest barrier heights of less than 60 kJ · mol -1 (specifically for 1,5[C ↔ N]-H and 1,6[C ↔ N]-H shifts). H-Migration barriers appear to decrease as the ring size in the transition structure (TS) increases, with a lowering of the barrier being found, for example when moving from a rearrangement proceeding via a four-membered-ring TS (e.g., the 1,3[C ↔ N]-H shift, CH 3-C(O)-NH . → .CH 2-C(O)-NH 2) to a rearrangement proceeding via a six-membered-ring TS (e.g., the 1,5[C ↔ N]-H shift, .NH-CH 2-C(O)-NH-CH 3 → NH 2-CH 2-C(O)-NH-CH 2 .).