Serum IgD levels were followed longitudinally twice a week for up to 100 days in 60 children undergoing allogeneic bone-marrow transplantation (n = 52) or immunosuppression (n = 8) for the treatment of leukaemia, severe aplastic anaemia or severe combined immunodeficiency. In 40 out of the 49 post-transplantation periods analysed (82%), a transient sharp increase of serum IgD was detected, irrespective of initial disease. A similar peak was found in one out of five children after immunosuppressive treatment. A second IgD peak was only recorded in grafted patients (14/49 post-transfusion periods). Peak levels of IgD ranged from 1.3 to 185.7 IU/ml (median 12.2 IU/ml), which represents a 2.6 to 22.4-fold increase over 'baseline' levels. In the transplanted leukaemia and aplastic anaemia patients, the rise of serum IgD occurred at the same time (geometric mean 16 days after transplantation) and was shown to represent heterogeneous polyclonal IgD in six of them. The onset of the serum IgD peak was significantly delayed in children suffering from severe combined immunodeficiency (P less than 0.05) and was demonstrated in one patient to consist of homogeneous IgD. No relation was found between either the occurrence of clinical acute graft-versus-host disease or infections after treatment, and the time of onset of IgD elevations. To detect transient serum IgD peaks as described here, frequent sampling of sera is necessary. The origin of the early IgD peaks seems to reside within the recipient's cells by an unknown mechanism. The late IgD peaks are most probably an expression of gradual reconstitution of the immune system following bone-marrow transplantation.