One of the obvious ways of testing the clonal selection hypothesis is to determine the minimum number of cells needed to initiate a measurable immune response. At least in the uncompromising, and now classical, hypothesis of one antigen-sensitive cell clone per antibody specificity (1, 2) it is mandatory that a given antigenic determinant should not be recognizable by every small lymphocyte--accepting that antigen-sensitive ceils are to be found among this, in several ways, heterogenous cell population. While this extreme result, if found, would truly invalidate the hypothesis it is unfortunately not possible to define exactly the maximum frequency of antigen-sensitive cells which would still be compatible with the hypothesis. Clearly, the more complex the immunological universe is thought to be, the higher becomes the number of different clones required by the hypothesis, and the lower becomes the expected frequency of a cell belonging to one particular clone. Few immunologists would probably expect a frequency higher than 10 -4.For reasons which were suggested by earlier studies of the factor of immunization (3), the strong histocompatibility antigens were considered to provide a particularly exacting test of the clonal selection hypothesis. The graftvs.-host (GVH) reactions in chicken embryos injected intravenously with adult chicken lymphocytes were used in the present study to determine the frequency of antigen-sensitive cells with re3pect to the strong antigens of the B locus (4). It is concluded that their frequency is too high to be compatible with the orthodox version of clonal selection.