Identification of wild animals that harbour the causative leptospires, and the identification of the most important of these 'wild reservoirs' (in terms of threat to human health), are key factors in the epidemiology of human leptospirosis. In an epidemiological investigation in the Australian state of Queensland, in 2007-2008, samples were collected from fruit bats (Pteropus conspicillatus) and rodents (to investigate the potential role of fruit bats in the maintenance and transmission of leptospires to ground-dwelling rodents) and checked for pathogenic leptospires. The results of these studies have now been carefully analysed in attempts to see which method of detection and type of test sample were best. The effects of pentobarbitone sodium used to euthanize wild mammals before collection of necropsy samples, on the survival and detection of leptospires in vitro, were also explored. In the earlier field investigation, serum, renal tissue and urine were collected from wild mammals, for the detection of pathogenic leptospires by culture, the microscopic agglutination test (MAT), real-time PCR and silver impregnation of smears. Although 27.6% of the rodents investigated were found leptospire-positive, culture only yielded four isolates, probably because many cultures were contaminated. The main aims of the present study were to quantify the performance of the individual diagnostic tests and examine the reasons behind the high incidence of culture contamination. The results of sensitivity and specificity analyses for the different diagnostic tests indicated that isolation by culture (the definitive diagnostic test for leptospiral shedding) had perfect (100%) sensitivity when compared with the results of the PCR but a low specificity (40%). The MAT performed poorly, with a sensitivity of 50% when compared against the results of culture. The prevalence of leptospiral carriage revealed by the PCR-based investigation of kidney and urine samples (59.2%) was higher than that revealed using any other method and far higher than the 2.0% revealed by culture. The results of the culture of renal tissue agreed fairly well with those of the PCR-based investigation of such tissue, with a Cohen's unweighted kappa coefficient (κ) of 0.5 (P = 0.04). The levels of agreement between other pairs of tests were generally poor. The presence of pentobarbitone sodium, at final concentrations of 27.8 or 167 mg/ml, did not affect the viability or the detection of leptospires in culture, and is therefore unlikely to reduce the chances of isolating leptospires from an animal that has been euthanized with the compound. It appears that collecting multiple samples from each mammal being checked will improve the chances of detecting leptospires (and reduce the chances of reporting an inconclusive result for any of the mammals). For the identification of a leptospiral carrier, however, the use of just two detection methods (culture and PCR) and one type of sample (renal tissue) may give adequate sensitivity and specificity. Given the robustness of PCR to contamination and its high sensitivity (it can give a positive result when DNA from just two leptospiral cells is present in the sample), a PCR-based serotyping method, to allow the combined detection and characterisation of leptospires from field isolates, would be extremely useful.