The main aim of this work is to illustrate some applications of Bayesian networks built for evaluating the value of DNA and criminalistics evidences. Some issues are presented concerning relatedness testing (paternity and identity investigation), coherent evaluation of different items of genetic evidence (obtained using additional markers – ChrY and mitochondrial polymorphism), phenotype predicting on the basis of knowledge of SNP polymorphism in some chosen genes and analysis of glass evidence. The developed networks may be easily extended when new information in a particular case is available and/or it is characterised by serious uncertainty.