It is generally assumed that the development of bacterial pneumonia becomes possible when the dose of inhaled or aspirated pathogens overwhelms the respiratory tract host defence system, but this hypothesis has not yet been tested either clinically or experimentally. This study evaluated inoculum dose in relation to onset of experimental pneumococcal pneumonia, and estimated the median effective dose resulting in pneumonia in healthy New Zealand White rabbits (mean+/-sd 4.75+/-0.25 kg (n = 27)). Rabbits were endobronchially inoculated with increasing doses of Streptococcus pneumoniae and pneumonia onset was observed over the following 96 h. The diagnostic approach was based on the Clinical Pulmonary Infection Score, modified for use in rabbits. Inoculation of S. pneumoniae at doses of >4.60 log(10) cfu made the development of pneumonia in rabbits more predictable (up to 90%). Lower doses of bacteria failed to cause pneumonia in 80% of inoculated animals. The median effective dose was estimated by means of logistical regression, probit analyses and the Reed-Muench method, and corresponded to 4.32, 4.38 and 4.67 log(10) cfu, respectively. It is speculated that development of pneumococcal pneumonia becomes more likely when the inoculum dose exceeds a threshold of antibacterial protection, making inoculum dose a risk factor for disease onset.