An infective basis for childhood leukaemia is not a new suspicion (Kel
lett, 1937). The failure of microbiologists to identify any specific a
gent and of epidemiologists to demonstrate marked space-time clusterin
g of the disease (Smith, 1382) have been discouraging, but neither is
incompatible with an infectious origin. In several vertebrate species,
the specific agents responsible for leukaemia belong to a class that
is notoriously difficult to isolate. Also, many infectious illnesses d
o not cluster because they are uncommon responses to the relevant infe
ction. Thus, the agent responsible for infectious mononucleosis is mai
nly spread not by those with the illness but by that very much larger
number of infected individuals who are clinically unaffected (or only
trivially so). Such infections can be considered as 'mainly immunising
': they can be seen as representing the most probable broad category t
o which the infection underlying childhood leukaemia belongs.