The workshops 2002-2008 showed an increasing degree of concordance of methods and applied systems. The observed errors involved only 0.1 percent - 0.3 percent of all PCR-based STR allele results. Currently, most discrepancies are due to clerical errors rather than differences in phenotype or nomenclature. The tendency of using the same commercial kits accounts for some of the uniformity in the results. Also, recommendations for nomenclature policy have contributed to improved consensus. Generally, there was high uniformity in the conclusions of the paternity testing exercises. The only discrepancy was in the 2005 exercise, in which the alleged father was closely related to the biological father. Still, the majority of the laboratories correctly excluded paternity and hypothesized a close genetic relationship between the biological and the alleged father. Most laboratories use computer programs for biostatistician calculations. Still, the paper challenges showed a significant variation in the formulas used by the participants, especially when rare alleles, inconsistencies, or silent alleles were present. Most laboratories consider the possibility of mutational events, but a large number of different formulas were used for calculating the likelihood ratio (LR), leading to significant variation in the total LRs. Silent alleles arise when the allele size is outside the range of measurement or when mutation in primer-binding site has occurred. The recommendation from the Paternity Testing Committee of the ISFG is to consider the possibility of silent alleles when only one allele is observed. 10 tables, 1 figure, 9 references, and appended listing of participating laboratories