Background: Over 20 years, from October 1989, the Darwin prospective melioidosis study has documented 540 cases from tropical Australia, providing new insights into epidemiology and the clinical spectrum. Principal Findings: The principal presentation was pneumonia in 278 (51%), genitourinary infection in 76 (14%), skin infection in 68 (13%), bacteremia without evident focus in 59 (11%), septic arthritis/osteomyelitis in 20 (4%) and neurological melioidosis in 14 (3%). 298 (55%) were bacteremic and 116 (21%) developed septic shock (58 fatal). Internal organ abscesses and secondary foci in lungs and/or joints were common. Prostatic abscesses occurred in 76 (20% of 372 males). 96 (18%) had occupational exposure to Burkholderia pseudomallei. 118 (22%) had a specific recreational or occupational incident considered the likely infecting event. 436 (81%) presented during the monsoonal wet season. The higher proportion with pneumonia in December to February supports the hypothesis of infection by inhalation during severe weather events. Recurrent melioidosis occurred in 29, mostly attributed to poor adherence to therapy. Mortality decreased from 30% in the first 5 years to 9% in the last five years (p<0.001). Risk factors for melioidosis included diabetes (39%), hazardous alcohol use (39%), chronic lung disease (26%) and chronic renal disease (12%). There was no identifiable risk factor in 20%. Of the 77 fatal cases (14%), 75 had at least one risk factor; the other 2 were elderly. On multivariate analysis of risk factors, age, location and season, the only independent predictors of mortality were the presence of at least one risk factor (OR 9.4; 95% CI 2.3-39) and age ≥ 50 years (OR 2.0; 95% CI 1.2-2.3). Conclusions: Melioidosis should be seen as an opportunistic infection that is unlikely to kill a healthy person, provided infection is diagnosed early and resources are available to provide appropriate antibiotics and critical care.