The effect of teat washing and drying on bacterial numbers in bulk milk was compared with that of no teat preparation in eight commercial herds over one year. Using in-line milk samplers, milk was collected at various points during its passage through the milking plant and the samples were used to establish the relative significance of the sources of contamination of raw milk. Teat washing and drying of cows housed during winter reduced the total counts by 40% and streptococcal and coliform counts by 50%. Bacterial counts were significantly lower in cows at pasture during the summer and there was no reduction in count due to teat washing and drying. Bacteriological counts increased at each stage as the milk passed through the milking machine. The milking equipment significantly increased the total colony count by between 2000 and 3000/ml, and the bulk tank added a further 1500 to 2000/ml. The mean rinse bacterial counts of the milking equipment were higher in summer than winter, averaging 4·4 x 107 bacteria/m2 compared with 3·5 x 107/m2 respectively. Although this level of bacterial contamination of the equipment is high by current standards, very low bulk milk bacterial counts were nevertheless achieved, particularly in the summer. This confirms that organisms from this source are not a major contaminant of the bulk milk. There was a very poor correlation between rinse counts and the bulk milk bacterial count, but a strong correlation (0·98) between total and streptococcal counts of the bulk milk. The unreliability of the use of rinse techniques to assess the contribution of milking equipment to bacterial counts of raw milk is emphasized.