Objectives: Fe-deficiency anaemia (IDA) occurs in 1-2 % of infants in developed countries, peaks at 1-3 years of age and is associated with later cognitive deficits. The objectives of the present study were to describe the characteristics of young children with severe IDA and examine modifiable risk factors in a developedcountry setting. Design: Two prospective samples: a national surveillance programme sample and a regional longitudinal study sample. Setting: Canada, 2009-2011. Subjects: Two samples of young children recruited from community-based healthcare practices: a national sample with severe anaemia (Hb < 80 g/l) due to Fe deficiency and a regional sample with non-anaemic Fe sufficiency. Results: Children with severe IDA (n 201, mean Hb 55·1 g/l) experienced substantial morbidity (including developmental delay, heart failure, cerebral thrombosis) and health-care utilization (including a 42 % hospitalization rate). Compared with children with Fe sufficiency (n 597, mean Hb 122·4 g/l), children with severe IDA consumed a larger volume of cow's milk daily (median 1065 ml v. 500 ml, P < 0·001) and were more likely to be using a bottle during the day (78 % v. 43 %, OR = 6·0; 95 % CI 4·0, 8·9) and also in bed (60 % v. 21 %, OR = 6·5; 95 % CI 4·4, 9·5). Conclusions: Severe IDA is associated with substantial morbidity and may be preventable. Three potentially modifiable feeding practices are associated with IDA: (i) cow's milk consumption greater than 500 ml/d; (ii) daytime bottle use beyond 12 months of age; and (iii) bottle use in bed. These feeding practices should be highlighted in future recommendations for public health and primarycare practitioners.