The performance of a variety of theoretical methods in computing stabilization energies of the substituted methyl and vinyl radicals ˙CH2F, ˙CH2CN, ˙CH2CHCH2, ˙CH2CHO, CH2 C˙F and CH2 C˙CN is examined. The influence of electron correlation (UHF, UMP2, PMP2, RMP2, UB3-LYP, UQCISD, UQCISD(T), UCCSD(T), URCCSD(T) and RRCCSD(T)) and basis set size (from 6-31G(d) to 6-311++G(3df,3pd)) on stabilization energies is evaluated, as well as the performance of compound methods such as G2, G3, CBS-Q and CBS-APNO and their variants. The results indicate that generally reliable radical stabilization energies can be obtained at modest cost using RMP2/6-311+G(2df,p)//RMP2/6-31G(d) energies. A slightly less accurate but more economical procedure is RMP2/6-311+G(d)//B3-LYP/6-31G(d). UMP2 and PMP2 are unsuitable for obtaining radical stabilization energies for spin-contaminated radicals, while UB3-LYP appears generally to overestimate stabilization energies.