We have re-evaluated mouse rosette formation (MRF) as a marker for B-CLL by estimating the fraction of mouse rosette forming B-lymphocytes (identified by CD20 monoclonal antibodies) in normal donors and malignant CD20+ cell proliferations (ALL, AML, B-NHL, B-HCL and B-CLL). Whereas this ratio was increased in B-CLL, all other CD20 positive malignancies showed mean ratios of less than 0.1. As part of a Danish multi-centre study, we furthermore prospectively analysed 86 patients and found that the mouse/CD20 ratio divided the 78 patients with monoclonal B-cell populations suspected of B-CLL in two distinct groups. In the low ratio group, three patients were categorized as leukaemized B-NHL and one as PLL. The remaining three patients with low ratio were clinically and immunologically (by SmIg density and CD5 expression) B-CLL patients suggesting a frequency of MR-negative B-CLLs of approximately 5%. In the high ratio group two of 70 patients were diagnosed as B-NHLs. Thirdly, MRF was a valuable parameter in patients, where transformation of disease is suspected, since it preceded clinical changes by several months. Thus, MRF is still a useful marker in the age of monoclonal antibodies.