False polymerase chain reaction (PCR) results may be obtained under unfavourable reaction conditions. Therefore optimal conditions for the different factors influencing a specific PCR method should be determined before introduction to a clinical diagnostic laboratory. This study has concentrated on the detection of heat-labile enterotoxin-producing E. coli by PCR, with empirical evaluation of various factors. Template was prepared by heat-lysis of E. coli, and this was shown to be adequate for PCR detection. The results showed that deviation from the optimal conditions of any of the following conditions may lead to false results: lysis of bacterial cells, denaturation temperature during cycling, annealing temperature, primer concentration, enzyme concentration, magnesium concentration and ion concentration. Three different detection methods for PCR product were also evaluated. As little as one bacterium can be detected after 35 cycles of PCR amplification with 32P-labelled oligonucleotide probe. An alkaline phosphatase-labelled probe was 10-fold less sensitive, whereas 100 bacteria in 10 microliters of the original sample suspension were necessary to give a positive signal after gel electrophoresis. The information in this study may be useful to those who wish to introduce PCR tests to diagnostic laboratories.