The responses of a random sample of consumers to the use of bovine somatotrophin (BST) in milk production were elicited using a stated preference methodology. A multinomial logit model of consumer choice was developed and tested to analyse consumers' choices of milk with varying characteristics of fat content, price, freshness and BST treatment. Welfare calculations for a representative consumer indicate welfare losses with the introduction of BST which are not fully offset by preferred milk attributes such as reduced price or increased freshness levels. Welfare losses were slightly less for a male than a female consumer and were less for consumers with higher levels of income and education. Losses were greatest when a representative consumer was denied the option of choosing not to purchase milk. There was a small welfare gain when the representative consumer was offered a full range of "BST" and "non-BST" milks. The results suggest that making appropriately labelled "BST-free" milk available to consumers could decrease consumer welfare losses associated with the introduction of BST in Canada.