This international case control study was conducted in 14 centers in 9 countries to investigate factors in childhood which may have a bearing on the etiology or pathogenesis of ulcerative colitis (UC) and Crohn's disease (CD). 197 patients with UC and 302 with CD (499 with inflammatory bowel disease (IBD] whose disease started before age 20 years and whose age at time of study was less than 25 years were investigated, with two age- and sex-matched controls for each patient. All subjects were studied with uniform questionnaires. Eczema was found significantly more frequently in patients with CD (p less than 0.005) and in their fathers (p less than 0.025), mothers (p less than 0.002), and siblings (p less than 0.01) as compared with their respective controls. IBD was significantly more frequent in parents, siblings, cousins, grandparents, and uncles of patients than in their respective controls. The fathers of patients with UC had significantly more major gastrointestinal and cardiovascular diseases at the time of the patient's birth than the fathers of controls. In North America mothers of patients with UC and CD took vitamin, mineral, and iron preparations during pregnancy significantly less frequently than mothers of controls. Patients with CD and UC consumed a lower residue diet than controls. Recurrent respiratory infections were more frequent in patients with UC and CD (p less than 0.001); it is uncertain whether this preceded disease. Hospitalization for respiratory diseases was more frequent in patients than controls, and the use of antibiotics more frequent in patients with CD. Smallpox vaccination was less frequent (p less than 0.05) in patients with CD, and chickenpox infection was less common in patients with UC (p less than 0.01). No significant differences were found between patients and controls in relation to various human and non-human contacts during childhood. Number of siblings, being an only child, and birth order did not differ markedly between patients and controls, and we could not confirm the 'sheltered child' hypothesis in IBD. The parents of controls were slightly better educated and their social class tended to be higher than those of parents of patients. There were significant associations between some of the main factors investigated in this study. No significant differences were found between patients and controls in the frequency of breast feeding, cereal consumption, sugar added to milk in infancy, gastroenteritis in childhood, major stressful life events, and many other factors.(ABSTRACT TRUNCATED AT 400 WORDS)