We present the results of the 1997, 1998 and 1999 Paternity Testing Workshops of the English Speaking Working Group of the International Society for Forensic Genetics. The numbers of participating laboratories were 24 (1997), 31 (1998) and 32 (1999). In 1997, all laboratories drew the correct conclusion that the alleged father was the biological father of the child. In 1998, the alleged father was the biological brother of the child and all laboratories excluded him. The scenario in 1999 was a deficiency case consisting of mother, child and the parents of the alleged father and all but one laboratory drew the correct conclusion.The percentage of laboratories routinely performing variable number of tandem repeat (VNTR) investigations using single locus probes (SLPs) and restriction fragment length polymorphism (RFLP) decreased from 83% in 1997 to 66% in 1999. In the three workshops, more than 90% of the laboratories used polymerase chain reaction (PCR) based systems. In 1999, 80% of the laboratories performing PCR, used commercially available short tandem repeat (STR) kits. Other commonly used systems were HLA and PolyMarker investigated with PCR. Conventional systems and RFLP investigations of VNTRs with multi loci probes (MLPs) were used routinely by approximately 20% of the participating laboratories.All laboratories submitting results in the three workshops used RFLP-based VNTRs or/and PCR based VNTRs/STRs. Inter-laboratory comparisons of the results showed a very high concordance. The overall coefficients of variation between the laboratories of the results of RFLP typing of the commonly used VNTRs D2S44, D7S21, D7S22 and D12S11 were 1.2-1.3%. Consistent results were obtained in the great majority of the systems investigated by PCR and typing errors counted for less than 0.3% of the PCR based results.