Objective: To investigate the molecular epidemiology of tuberculosis, temporal and spatial distribution of Mycobacterium tuberculosis isolates and associations between genotypes and clinical characteristics, in a low prevalence population. Methods: A total of 930 M. tuberculosis isolates referred to the New South Wales (NSW, Australia) Mycobacterium Reference Laboratory in 2004-2006 were characterised by mycobacterial interspersed repetitive unit (MIRU) and spacer oligonucleotide (spoligo) typing. Associations between genotypes, patient age, disease site and drug resistance were explored and the predictive power of molecular typing was analysed using Bayesian Belief Networks. Results: Among isolates from 855 NSW residents, there were 287 spoligotypes, 494 MIRU types and 643 unique spoligotype-MIRU type combinations. They formed 73 spoligotype, 104 MIRU type and 76 spoligo-MIRU clusters, most of which contained only two isolates. The majority (87.7%) of spoligotype clusters contained several MIRU profiles and 64.4% of MIRU clusters contained several spoligotypes. The three most common M. tuberculosis clades were Beijing (24.1%), East African Indian (11.8%) and Central Asian (6.5%); 6.9% and 0.7% isolates were resistant to isoniazid and rifampicin, respectively. There was no proof of association between genotype and drug resistance but isoniazid resistance increased independently over time. Given the low rates of genotype clustering, statistical analysis of genotype-phenotype associations was limited. Potential associations were not confirmed by Bayesian classifiers. Conclusions: Spoligo and MIRU typing demonstrated low levels of M. tuberculosis clustering in NSW; temporal and spatial changes in M. tuberculosis genotypes reflected migration patterns to Australia. No analytically significant associations between M. tuberculosis genotypes and clinical phenotypes were detected.