Splenic T cells exposed to concanavalin A (Con A), and subsequently to factors produced by rat spleen cells in response to Con A (Con A sup), acquire the ability to function as helper T (TH) cells in response to xenogeneic erythrocytes (RBC). Help is measured as the reconstitution of the plaque-forming cell response of a spleen cell population depleted of T cells by treatment with anti-Thy-1 serum and complement. We propose that precursor TH cells differentiate during the in vitro treatment into mature TH cells. As differentiation occurs under limiting dilution conditions, an estimation of the precursor frequency should in principle be possible. However, a single-hit Poisson distribution does not fit our data. Instead, we observe, dependent on the T cell concentration, three separate "peaks" of response. In many experiments, using sheep, horse, and chicken RBC as antigens, we reproducibly find these "peaks" at 40-190, 600-3,000, and 20,000-100,000 T cells, placed into limiting dilution cultures, respectively. By various experiments we can show that the helper activity is not due to passively transferred rat factors, but to the titrated cells themselves. The active cell is a T cell that appears to function in an antigen-specific way and to require direct cell contact to do so. It thus resembles the classical helper T cell. As we find precursor TH cells already at very low concentrations of T cells, we titrated the range between 0 and 100 T cells/well carefully. The bent shape of the titration curves does not always allow a statistically satisfying regression analysis, and we therefore cannot estimate precise precursor frequencies from every experiment. However, a common sense argument can be made that these frequencies must be on the order of 1/10-1/100 T cells. We propose that the limiting dilution curves obtained in this system most likely reflect fundamentally important cellular interactions that regulate immunological effector functions. We favor a concept of independently interacting sets of helper and suppressor T cells of various frequencies, but other models are possible.