Two hundred and seventy-four mothers who were bottle feeding their infants aged from 1 to 9 months, were interviewed. Particular attention was paid to mixing technique, storage of reconstituted formulae, the reasons for deciding to bottle feed and what influenced the choosing of a particular milk. Following the interview, a sample of milk from a previously prepared bottle was taken to measure osmolality and bacterial colony count. Eighty-two per cent of the mothers had initially attempted breast feeding. The commonest reason for changing to a bottle was that the mother felt she was not able to produce enough milk to satisfy the baby. Thirty per cent chose the milk they were using on the recommendation of the maternity hospital. Errors in reconstituting the formulae, compared with the manufacturer's instructions were made by 100 (30%) mothers. In 52 cases these were potentially serious errors, usually erring on the side of preparing an over-concentrated formula. This finding from the interviews was confirmed by osmolality analysis of milk samples. Twenty-two per cent of samples collected grew potential pathogens. There was no correlation between the presence of bacterial growth and the use of prewarmed formulae stored in insulated carriers. No correlation was found between triceps skinfold thickness, age of introduction of solids or method of reconstitution of formula.