The in vivo and in vitro effects of GM-CSF were tested in four patients with large granular lymphocyte proliferation (LGLP) and severe granulocytopenia. All patients had an increased percentage of LGL cells (greater than 20%), whereas 3/4 patients demonstrated rearranged T-cell-receptor genes. An effect on the peripheral granulocyte counts was noticed in 3/4 patients after 14 days of GM-CSF administration (5 micrograms/kg/day, subcutaneously); 2.5, 7, and 0.45 x 10(9)/l, respectively. These changes were associated with a two- to five-fold increase in monocytes and a strong increment in eosinophils. In one patient GM-CSF was not effective in increasing the granulocyte count. B lymphocytes showed a variable increase, ranging from 1.2-fold to 3.8-fold, while the number of NK cells remained almost constant during the GM-CSF treatment. No consistent effect of GM-CSF on T lymphocytes was observed. These data suggest that GM-CSF may be a therapeutic option in some patients with LGLP complicated by granulocytopenia and/or infection.