To investigate potential routes of spread of infection by the polymerase chain reaction (PCR) it is important that the technique is effective in the types of specimen to be investigated. To establish the limits of detection of Helicobacter pylori by PCR in clinical material from the gastric mucosa, faeces, dental plaque and oral rinses, samples were seeded with known numbers of bacteria. DNA extraction was followed by amplification with primers from the urease C gene. Nested primers were used to amplify the PCR product which was detected using a digoxigenin-labelled probe. Faeces or plaque inhibited the single reaction 10(2)-10(6) fold. A second amplification using nested primers and probing increased the sensitivity to a level similar to that obtained with pure culture. This method is potentially useful with less likelihood of false negative results when trying to detect H. pylori by PCR in highly contaminated, clinical material.