Laboratory models enabling to study in vivo human leukocyte functions have been developed. Most of the models consist of human immunocytes transferred to mice homozygous for the scid mutation. Mice with additional immunodeficient-prone genetic background or with immunodeficiency-induced conditioning have also been used. Human grafts mainly consisted of human immune cells in suspension injected intraperitoneally, or in pieces of human organs containing immunocytes implanted subcutaneously. Cells in suspension could be easily manipulated in vitro before transfer to the animal, but disseminated within the mouse body. In opposition, human cells mostly remained within implantation areas of animals given human organ pieces. This favorizes cell interactions and helps for cell recovery after their in vivo passage. Moreover, the diversity of antibodies in animals transplanted with human lymphoid organ pieces appeared broader than that of mice transferred with lymphocytes in suspension. Spontaneous recall antibody and autoantibody productions have been generally observed in animals transferred with cells from donors with such antibodies. In vivo boosting of recall antibody by antigen has been most successful, but such a manipulation inconstantly boosted autoantibodies. Primary human T and B cell responses were difficult to obtain in xenochimeric animals, and success has been generally obtained by optimizing human immune response parameters, such as antigen presentation.