All newborn infants in East Anglia are screened for cystic fibrosis by blood immunoreactive trypsin assay at 7 days. Thirty eight infants with cystic fibrosis were randomised to treatment with either continuous oral flucloxacillin 250 mg/day (group P, n = 18) or with episodic antimicrobials as clinically indicated (group E, n = 20). Their progress was monitored from diagnosis to 24 months by a nurse coordinator who visited all infants regularly, at home and in hospital, to collect anthropometric, dietary, clinical, and microbiological data. Mean (range) age of confirmation of diagnosis was 5.7 weeks (1-14 weeks). There was no significant difference in birth weight, genotype, immunoreactive trypsin concentration, neonatal history, symptoms at diagnosis, pancreatic enzyme supplementation, or parental smoking history between the groups. Infants in group E had more frequent cough and a greater number of Staphylococcus aureus isolates than infants in group P. More infants of group E were admitted to hospital, had higher admission rates during the second year (19 v 5), for longer periods (6.4 v 2.2 days), despite receiving more than double the number of courses of antibiotics than group P infants (in addition to flucloxacillin). Continuous prophylactic flucloxacillin from early diagnosis of cystic fibrosis is associated with improved clinical progress during the first two years of life.