Cartons and bottles of whole pasteurized cows milk were purchased from retail outlets throughout central and Southern England and south Wales month-by-month from 1st September 1991 to 31st March 1993. These were brought into a dedicated unit in the department and processed under experimental conditions which rigorously excluded contamination artifact, monitored by simultaneous process controls. Each container was cleaned with 100% ethanol, a 15 ml aliquot of milk aspirated and centrifuged 41,000xg for 1 hr yielding cream, almost clear whey and pellet fractions. These were separated and made up to the volume of the whey. 500 mcl aliquotes of each fraction were then boiled for 20 min, centrifuged and 5 mcl of supernatant assayed in triplicate by IS900 PCR, with individual PCR negative and positive controls. The lower limit of detection of this system, previously established in spiked samples, lay in the range 500-1000 M.ptb organisms per ml. The results showed that coming in surges particularly in Feb/Mar and Sept/Oct in which up to 1/4 of samples were affected, an overall 6.25% (21 of 336) of cartons or bottles tested positive by IS900 PCR. In 18 of 21 positives (86%) the PCR signal was present in cream and/or centrifugal pellet which is where the PCR signal segregates in milk spiked with intact M.ptb. 575 liquid culture flasks derived from 19 PCR positive milks, 34 PCR negative milks and multiple buffer and medium-only blanks currently incubated for between 8 and 26 months, await meticulous testing for viable M.ptb. The results to date are consistent with an intermittent contamination of pasteurized cows milk supplies by intact M.ptb.