Background: Latency is a key feature of animal pathogen Chlamydia abortus where infection remains inapparent in the non-pregnant animal and only becomes evident during a subsequent pregnancy. Often the first sign that an animal is infected is abortion occurring late in gestation. Despite this, little is understood of the underlying mechanisms that control latency or the recrudescence of infection that occurs during subsequent pregnancy. The aim of this study was to develop an experimental model of latency by mimicking the natural route of infection through the intranasal inoculation of non-pregnant sheep with C. abortus. Methodology/Principal Findings: Three groups of sheep (groups 1, 2 and 3) were experimentally infected with different doses of C. abortus (5 x 10^33, 5 x 10^5 and 5 x 10^7 inclusion forming units (IFU), respectively) prior to mating and monitored over 2 breeding cycles for clinical, microbiological, pathological, immunological and serological outcomes. Two further groups received either negative control inoculum (group 4) or were inoculated subcutaneously on day 70 of gestation with 2 x 10^6 IFU C. abortus (group 5). Animals in groups 1, 2 and 5 experienced an abortion rate of 50-67%, while only one animal aborted in group 3 and none in group 4. Pathological, microbiological, immunological and serological analyses support the view that the maternal protective immune response is influenced by initial exposure to the bacterium. Conclusions/Significance: The results show that intranasal administration of non-pregnant sheep with a low/medium dose of C. abortus results in a latent infection that leads in a subsequent pregnancy to infection of the placenta and abortion. In contrast a high dose stimulates protective immunity, resulting in a much lower abortion rate. This model will be useful in understanding the mechanisms of infection underlying latency and onset of disease, as well as in the development of novel therapeutics and vaccines for controlling infection.