Lyme borreliosis caused by the spirochete Borrelia burgdorferi is now the most common vectorborne disease in North America, Europe and Asia. It is a multisystemic infection which may cause skin, neurological, cardiac or rheumatologic disorders. The aims of the present thesis were: (i) to develop a PCR assay for direct detection of B. burgdorferi DNA and to evaluate the diagnostic utility of PCR in clinical specimens from patients with Lyme borreliosis and (ii) to study the taxonomic classification of B. burgdorferi isolates and its implications for epidemiology and clinical presentation. Laboratory diagnosis of Lyme borreliosis by direct demonstration of B. burgdorferi in clinical specimens would compared to current serology allow (i) optimal specificity, (ii) increased sensitivity during the first weeks of infection, when the antibody response is not yet detectable and (iii) discrimination between ongoing and past infection. Due to the extreme paucity of spirochetes in clinical specimens neither in vitro culture nor antigen detection had yielded a sufficient diagnostic sensitivity. Thus the recently introduced highly sensitive PCR methodology could be a solution and was thus studied. Assays for PCR amplification and subsequent identification of B. burgdorferi specific sequences were established and used. For all assays the analytical sensitivity was a few genome copies using purified DNA as template. The efficacy of PCR was initially evaluated using tissue samples from experimentally infected gerbils in order to start with biological samples a priori known to contain B. burgdorferi. B. burgdorferi DNA was detectable in 88% of the specimens. Thus the diagnostic sensitivity of PCR was comparable to and even higher than in vitro culture. PCR was significantly more sensitive than a histological B. burgdorferi specific immunophosphatase-staining method. The utility of the PCR was then tested for identification of B. burgdorferi DNA in skin biopsies from 31 patients with erythema migrans. The sensitivity of PCR was 71%, which was superior to culture and serology. Based on own and otherwise published results there is clear evidence for PCR being the most sensitive and specific test for detection of B. burgdorferi in skin biopsies from patients with both early and late dermatoborreliosis. However, since the clinical diagnosis of dermatoborreliosis in most instances is easy, an invasive procedure as a skin biopsy, will only be justified in patients with an atypical clinical presentation. The most frequent and serious manifestation of disseminated Lyme borreliosis is neuroborreliosis. PCR was applied to 190 patients with untreated and confirmed neuroborreliosis. B. burgdorferi DNA was detectable in 17-21% of CSF samples from patients with neuroborreliosis. In patients with very early neuroborreliosis (< 2 weeks), still being negative for specific intrathecal antibody synthesis, a positive PCR was more frequent than in patients with longer disease duration. PCR can be used as a diagnostic aid in these patients. However, in general the measurement of specific intrathecal antibody production in patients with neuroborreliosis was superior to PCR. In urine samples from patients with Lyme borreliosis the diagnostic sensitivity varied, generally showing a low reproducibility. Urine is thus not regarded as a suitable sample source for B. burgdorferi PCR. The reason may be the variable presence of Taq polymerase inhibitors. Based on a semi-quantitative detection system for amplicons, reflecting the input amount of specific DNA and thus the density of spirochetes in the clinical samples high amounts of DNA were found in skin biopsies whereas especially in urine the amount of DNA was low. When the present study was initiated there was no accepted classification of B. burgdorferi. A heterogeneity among B. burgdorferi strains might have important implications for understanding the epidemiology and different clinical presentations (dermatoborreliosis versus neuroborreliosis) and courses (self-limiting versus chronic disease). Furthermore, strain differences were of importance for selection of suitable antigens for diagnostic assays and for vaccine development. Since then, B. burgdorferi isolates have been studied by phenotypic and genotypic traits and have been shown to be highly heterogeneous. Our first approach was to genotype a panel of human B. burgdorferi isolates by restriction fragment length polymorphism (RFLP) of three genes. Thereafter, sequencing and dideoxy fingerprinting of ospA was applied. By RFLP the strains could be differentiated into two to five groups. The RFLP classification was compared with four different phenotypic and genotypic methods including the rRNA typing. Results obtained with the different methods correlated highly and confirmed the meanwhile accepted taxonomic classification by Baranton et al., According to this the term B. burgdorferi sensu lato comprises three different human pathogenic genospecies B. burgdorferi sensu stricto, B. garinii and B. afzelii. All three genospecies have been isolated among Danish patients with Lyme borreliosis and are thus prevalent in Denmark. Since isolation of B. burgdorferi from patients with Lyme borreliosis is laborious and often unsuccessful molecular typing methods based on PCR are recommended obviating the need for isolation by prior culture. Of special interest was to study a possible association of neuroborreliosis to certain B. burgdorferi genospecies, indicating species depended organotropism. By RFLP all six CSF isolates tested belonged to B. garinii and that 6 out of 7 isolates from patients with acrodermatitis chronica atrophicans belonged to B. afzelii. Due to the low culture yield of B. burgdorferi from CSF, the association of B. garinii and neuroborreliosis was further studied by sequence analysis and dideoxyfingerprinting analysis of ospA PCR amplicons obtained from CSF samples from patients with neuroborreliosis. Phylogenetic analysis showed that in 11 out of 13 patients B. garinii DNA was found in CSF. These data strongly supports the hypothesis that B. garinii is the principal agent of Lyme neuroborreliosis in Europe. Similarly it was shown that B. afzelii is associated with acrodermatitis chronica atrophicans and thus dermatoborreliosis. Due to a strain dependent different selection pressure in culture only PCR based methods can be used to answer whether mixed infection in patients specimens occur. Our data indicate that mixed infections in humans if ever are rare.