The paper attempts to analyze quality and safety attributes of milk for which consumers are willing to pay and make their decisions to purchase using the framework of conjoint analysis. The study also examines consumer’s level of preference concerning quality and safety attributes of milk in relation to their socio-economic factors. The data used for this analysis is collected from a consumer survey of 900 households conducted in April 2006 and a supplementary survey specifically designed for the conjoint study with a sub-sample of 380 households from the original 900 samples- 260 in Dhaka city and 120 in Mymensingh city of Bangladesh, conducted in 2008. Buyers of raw fresh milk appeared to be fairly indifferent about level of fat content, have marginally more preference for milk from local breeds rather than crossbreeds but have strong preference for milk in which water has not been added. Low price is also stronly preferred to high price, which is an indication that consumers do not associate high price with higher quality, given that several other attributes such as breed of the cow and adulteration are not easily verifiable at the point of purchase. Water addition is the most common form of adulteration of milk in the country. Consumers perceive that water addition has two potential negative effects on the quality of milk. First, addition of water may make the milk ‘impure’ or ‘unsafe’ if contaminated water is addeded, and it dilutes milk so fat content is reduced in water added milk. Therefore, preference for fat content is partly indirectly expressed through preference for milk without added water. There are other forms of adulteration such as adding powdered milk, chemicals to avoid spoilage but ordinary consumers can’t easily verify these at the time of purchase to make a choice. Buyers of pasteurized milk are also indifferent about fat content but have strong preference for milk in polypack rather than in paper cartoon and for low price. It was found that the consumers gave highest consideration on freshness of raw milk followed by taste and purity. In case of pasteurized milk, taste was the most important criteria, and fat content the least. It appeared from conjoint analysis that, other things being equal, the most preferred profile of raw fresh milk is ‘milk from local breed cow with low fat without water adulteration and at low price’ and the least preferred is ‘milk from a crossbred cow with low fat, added water and at high price’. In case of pasteurized milk, the most preferred profile is ‘full cream milk in polypack at low price’ and the least preferred profile is ‘low fat milk in paper cartoon at high price’. Among the selected attributes of raw fresh milk, other things being equal, an average buyer in his/her purchase decision give 31% weight on price, 28% on water adulteration, 22% on breed of the cow and 19% on fat content. In case of pasteurized milk, among the three selected attributes 49% of weight is given on price, 30% on packing and 21% on fat content. Regression analysis showed that fat content, cow breed, milk purity, milk price, religion, household expenditure per month and district dummy significantly affected product profile preference rating of raw milk. The result of this study could be used for designing safety and quality standard of milk for domestic market and gradually update those standards as new information on quality criteria and consumer preferences emerge from new research.