An international collaborative study was performed to evaluate a set of PCR reference reagents for HIV diagnosis. Twenty-six laboratories from 9 countries analysed a proficiency panel of 10 coded DNA samples using the PCR reference reagents and protocols. For comparison, these coded samples were then assessed using a laboratory's own 'in-house' reagents and methodologies. The objectives of the study were: (i) to assess inter-laboratory variation of PCR sensitivity, (ii) to evaluate the DNA 'carryover' problem and frequency of false negative results and (iii) to examine the utility of the complete set of reagents and templates to act as reference preparations for HIV PCR. Using the reference reagents, 46% of laboratories reported no false positive results in any of their assays of the negative controls. The remaining laboratories all reported a false positive result(s) in at least one assay. The overall false positive result rate for the study was 9.3%. In contrast, an overall false negative result rate of 7.4% was observed, with some laboratories recording negative results even for samples containing 10,000 molecules of target DNA. The level of absolute sensitivity may be assessed accurately only from the 12 laboratories that obtained no false positive results. All 12 laboratories detected the sample containing 10 molecules of template DNA and 9 out of the 12 laboratories detected the sample containing 1 molecule. This is in close agreement with the theoretical detection rate based on a statistical probability model for the detection of a single molecule. These characterised reference reagents were at least as sensitive as any of the 'in-house' reagents and methodologies applied, including nested PCR. The complete set of characterised reference reagents is now available for quality control assessment of HIV-1 PCR from the MRC ADP.