Cross-species transfers of pathogens (zoonoses) cause some of the most viru
lent diseases, including anthrax, hantavirus and Q fever. Zoonotic infectio
ns occur when a pathogen moves from its reservoir host species into a secon
dary host species. Similarly, commensal infections often have a primary res
ervoir location within their hosts' bodies from which they rarely cause dis
ease symptoms, but commensals such as Neisseria meningitidis cause severe d
isease when they cross into a different body compartment from their normal
location. Both zoonotic and commensal infections cause either mild symptoms
or severe disease, but rarely intermediate symptoms. We develop a mathemat
ical model for studying three factors that affect the probability of severe
disease: the size of the inoculum, the route of inoculation and the freque
ncy of naturally occurring infections that do not cause symptoms but do ind
uce protective immunity (vaccinating inoculations). With a single route of
infection, increasing pathogen density causes inoculations to develop more
often into disease rather than asymptomatic vaccinations that provide prote
ctive immunity. With two routes of infection, it may happen that a lower de
nsity of a pathogen or of a particular antigenic variant leads to a relativ
ely higher frequency of disease-inducing versus vaccinating inoculations. T
his reversal occurs when one route of infection tends to vaccinate against
relatively common pathogens but less often vaccinates against relatively ra
re pathogens, whereas the other route of infection is susceptible to diseas
e-inducing inoculation even at relatively low pathogen density.