The risk profile of childhood leukaemia in Greece was studied through a case-control investigation that included all 153 incident cases of the disease, ascertained throughout the country during 1993 and 1994, and two hospital controls for every case matched for gender, age and place of residence. The data were analysed using conditional logistic regression and the associations are expressed in terms of adjusted odds ratios (OR) and their 95% confidence intervals. Cases were born to mothers of a higher standard education, the OR for an increment of four schooling years being 1.48 (1.17-1.87) and had higher birth weight, the OR for an increment of 500g being 1.36 (1.04-1.77). Pet ownership and birth after a pregnancy with anaemia were associated with increased risk, the ORs being 2.18 (1.14-4.16) and 2.60 (1.39-4.86) respectively. From the frequency analyses, indicative inverse associations were found with birth order, household crowding and previous hospitalization with allergic diseases, whereas indicative positive associations were found with diabetes mellitus during pregnancy and with neonatal jaundice. Substantial or significant elevations were not found with respect to maternal smoking and coffee drinking during pregnancy, diagnostic radiography and ultrasonographic examinations or blood transfusions. A significant inverse association with maternal consumption of alcohol could be due to multiple comparisons, but a detrimental effect can probably be excluded. A non-significant positive association with total shots of viral vaccinations and a weak non-significant inverse association with breast feeding were also found. We interpret the findings of this study as being compatible with acute childhood leukaemia being linked with delayed development of herd immunity to fairly common infectious agents, in conjunction with accelerated perinatal and early post-natal growth.