The presence of pathogens must be adequately monitored in raw milk and dairy products to assess their safety. With this aim, sampling plans and analytical methods should be evaluated for their appropriateness. Five incremental samples were collected from 40 milk tankers and overall 200 individual samples were analysed to assess the number and distribution of Escherichia coli O157 and O26. Enriched samples were screened by PCR for the presence of the serogroup-specific genes rfbEO157 and wzxO26, and the positive cultures were analysed by reverse transcriptase-PCR to detect mRNA sequences encoded by these genes, which are indicative of the presence of viable bacteria. In addition, the numbers of E. coli O157 and O26 in milk were estimated by most-probable-number–PCR method. The results of PCR screening targeting the rfbEO157 and wzxO26 genes were positive in 34 samples (representing 15 batches) and complementary mRNA sequences were detected in 27 of these (80 %). Isolation from the enriched cultures resulted in detection of E. coli O157 in 100% of the enriched samples positive for the O157 RNA marker, whereas E. coli O26 was isolated from only 37.5 % of the enriched cultures positive for O26 RNA marker. E. coli strains that produce Shiga toxins O26:H11 was isolated from one batch (2.5 %), while typical enteropathogenic E. coli O26 and non-pathogenic strains of E. coli O157 were isolated from four other batches. Estimated by most-probable-number (MPN), the number of E. coli O157 or O26 was below 0.3 MPN ml−1 in all the milk samples, except two that had MPN equal to 1.4 ml−1 (CI95% 0.36–4.2). The distribution of E. coli O157 and O26 within the milk batches was not uniform. Seven (47 %) of the 15 batches with positive results for either the O157 or O26 gene had only one incremental sample positive out of five samples analysed. In the presence of very low concentrations of pathogens, the sampling method used may affect the probability of detection of ‘defective’ batches.