The reproducibility of random amplified polymorphic DNA (RAPD) was tested using two different thermal cyclers and three brands of Taq DNA polymerase. Three different oligonucleotides were used to obtain patterns of amplified fragments from three DNA samples (Escherichia coli, Bacillus subtilis, and Thermococcus littoralis). Experiments were repeated three to six times. Apart from the expected between-oligonucleotide and between-DNA variations, between-thermal cycler and between-DNA polymerase variations were observed. Within the DNA-oligonucleotide-DNA-polymerase-thermal cycler, reproducibility was excellent when the thermal cycler equipped with the best temperature regulation was used, but was not as good with another brand of thermal cycler.