We performed personal identification by DNA polymorphism using forensic specimens. DNA was purified by the potassium iodine method instead of the phenol extraction method. When DNA is amplified by PCR, there are inhibitors of PCR such as melanin in hairs or blood and inorganic salts in bone. This problem was solved by the potassium iodine method. The genotype of the ABO blood groups could be determined by the polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP) method in all 29 old blood-stain specimens (obtained 11 months-15 years before) and 14 fingerprint specimens on a glass plate. In sexual assaults against women, one key to identifying the suspect is ABO phenotyping or the typing of other polymorphic markers of the seminal fluid in the victim's vagina. However, ABO phenotyping is frequently unsuccessful, since mixtures of fluids cannot be separated to be subjected to conventional methods for the detection of antibody or antigen material. We therefore studied ABO blood group genotyping of sperm DNA isolated from contaminating vaginal fluid by the PCR-RFLP method. Seminal samples of genotypes OO, AO, BO and AB were experimentally mixed with vaginal fluid (OO, AO, BO and AB), and were successfully separated and genotyped by this method. In practice, we also separated and genotyped the seminal DNA of suspects from contaminated postcoital vaginal fluid obtained in 4 sexual assaults. These forensic samples were easily separated and completely genotyped. This reliable ABO genotyping method by PCR-RFLP, using separated sperm DNA, should be of value in forensic identification in sexual assaults. We attempted sex and individual identifications of several forensic specimens by detecting various sex chromosome-specific sequences by the polymerase chain reaction (PCR) method. The specimens were 30 blood stains that were attached to gauze and stored 11 months-15 years (15 males, 14 females and 1 of unknown gender) and 10 bleached white bones (8-about 10000 years after death). Of the known 29 blood stains, the sex determination rate was the highest by DYZ-1 (amplified DNA fragment = 149bp, about 3000 copies on Y chromosome) recombinated with DXZ-1 (28 of the 29 samples) although 27 samples could be determined by DYZ-1 (amplified DNA fragment = 1000bp, about 3000 copies on Y chromosome) recombinated with DXZ-1. The sex determination rate was relatively low by DYZ-3 (amplified DNA fragment = 170bp, about 100 copies on Y chromosome) recombinated with DXZ-1 (23 of the 29 samples). The sex determination rate by two single locus markers, PAB or Amelogenin was markedly low (7 and 9 of 29 samples, respectively). The sex determination rate by the 27H39 locus was 27 of the 29 samples; 14 of the 15 male were determined (A = 186bp, 1; B = 190bp, 7; C = 194bp, 5; D = 198bp, 1; E = 202bp, 0). One unknown gender could be identified as male by multiple locus markers, but could not be identified by any single locus markers except for the 27H39 locus. Of the 10 bleached white bone specimens, 9 could be determined by the combination of DYZ-1 and DXZ-1, but the results of determination were inconsistent with the anthropological characteristics in 3 cases. The 27H39 locus showed polymorphism in 4 of the 6 anthropologically male beached bones. Studies on a new microsatellite on the Y chromosome are needed for evaluation of polymorphism on the Y chromosome because it is not necessary to consider the alleles from the female in the paternity test in male children or female-derived elements in sexual delict, and there is a distance between the alleles using 27H39 alone. In this study, We evaluated the allele distribution of DYS384, DYS388, DYS389, DYS390, DYS391, DYS392, and DYS393, in Japanese subjects and found the usefulness of DYS384 and DYS390. Future studies on personal identification by DNA polymorephism will mainly evaluate short tandem repeat (STR).