Isotropic hyperfine couplings have been calculated and compared to experiment for 25 radicals containing phosphorus using the 6-311G** basis set. At HF/6-31G** optimized geometries, Fermi contact integrals derived from UMP2 spin-density matrices give the best correlation with experiment (20 data points for P-31, 8 data points for F-19, 7 data points for Cl-35, and 5 data points for H-1 hyperfine couplings, rms error of 22.9 G with the data spanning a range of about 1650 G); PUHF spin-density matrices are somewhat less useful (rms error of 33.6 G), and calculation at the UHF level gives poor results (rms error of 45.6 G). Hyperfine couplings calculated at the UMP2 level may be scaled by 102% to bring them into closer agreement with experiment. Even though geometric changes on going to the MP2 level may be sizable, better correlation with experiment is obtained when HF/6-31G** optimized geometries are used instead of MP2/6-31G** optimized geometries (rms error of 36.1 G). Including diffuse functions in the calculation of the hyperfine couplings for anionic radicals has in general only marginal effect. Statistical trends within specific subsets of the data are also discussed.