A paternity case is presented in which DNA typing with thirteen variable number of tandem repeat (VNTR) systems was in apparent conflict with an initial multilocus DNA analysis using oligonucleotide probe (CAC)5. A dual approach was necessary, because the multilocus analysis yielded a proportion of aberrant nonmaternal offspring bands far too small and level of band-sharing too high for non-paternity. The case could finally be solved by reference to the joint likelihood ratio combining both approaches. Arguments are presented that summation of log likelihoods is indeed valid even if no information is available regarding linkage between the VNTR loci and the loci contributing to a multilocus DNA fingerprint.