When using a standard battery of STRs for relationship testing a small proportion of analyses can give ambiguous results - where the claimed relationship cannot be confirmed by a high enough paternity index or excluded with fully incompatible genotypes. The majority of such cases arise from unknowingly testing a brother of the true father and observing only a small number of exclusions that can each be interpreted as one- or two-step mutations. Although adding extra STRs might resolve a proportion of cases, there are few properly validated extra STRs available, while the commonly added hypervariable SE33 locus is four times more mutable than average, increasing the risk of ambiguous results. We have found SNPs in large multiplexes are much more informative for both low initial probabilities or ambiguous exclusions and at the same time provide a more reliable genotyping approach for the highly degraded DNA encountered in many identification cases. Eight relationship cases are outlined where the addition of SNP data resolved analyses that had remained ambiguous even with extended STR typing. In addition we have made simulations to ascertain the frequency of failing to obtain exclusions or conclusive probabilities of paternity with different marker sets when a brother of the true father is tested. Results indicate that SNPs are statistically more efficient than STRs in resolving cases that distinguish first-degree relatives in deficient pedigrees.