Coupled-state calculations including positronium channels are reported for positron scattering by atomic hydrogen, lithium and sodium. Integrated cross sections and total cross sections are presented for all three atoms. For lithium differential cross sections are also given. Throughout, comparison is made between results calculated with and without inclusion of the positronium channels. S-wave cross sections for positron scattering by atomic hydrogen in the Ps(1s, 2s, 2p)+H(1s, 2s, 2p) approximation show the high energy resonance first observed by Higgins and Burke in the coupled-static approximation. This resonance has now moved up to 51.05 eV and narrowed in width to 2.92 eV. Other pronounced structure is seen in the S-wave cross sections between 10 and 20 eV; it is tentatively suggested that this structure may be due to the formation of a temporary pseudo-molecular collision complex. Results calculated in the Ps(1s, 2s, $$\overline {3s} ,\overline {4s} $$ , 2p, $$\overline {3p} ,\overline {4p} ,\overline {3d} ,\overline {4d} $$ ,+H(1s, 2s, $$\overline {3s} ,\overline {4s} $$ , 2p, $$\overline {3p} ,\overline {4p} ,\overline {3d} ,\overline {4d} $$ approximation show convergence towards accurate values in the energy region below and in the Ore gap. Contrary to previous work on lithium using only an atomic basis, it is found that coupling to the 3d state of lithium is not so important when positronium channels are included; this is because a mixed basis of atom and positronium states gives a more rapidly convergent approximation than an expansion based on atom states alone. The threshold behaviour of the elastic cross section and the Ps(1s) formation cross section for lithium is investigated. Results in the Ps(1s, 2s, 2p)+Na(3s, 3p) approximation for sodium show good agreement with the total cross section measurements of Kwan et al.