Infection of vectors is often assumed to depend on a threshold level of viremia. In this model, all vectors which bite a viremic host become infected if the viremia is above the threshold and do not become infected if the viremia is below the threshold. This is an oversimplified view of a process which may have complex relationships between virema, host survival, probability of infection during a single bite, and number of vectors which bite during the course of the viremia. We examined the probability of infection and number of infected vectors under different assumptions about viremia and infection. When infection depends on a threshold number of virions ingested in the blood meal, the probability of infection declines exponentially as the threshold increases. If, however, infection can occur with only one virion, infection can occur even at low viremias. If high viremias result in host death, under the threshold model the maximum number of vectors will be infected by hosts with viremias at or just above the threshold needed for infection. When viremia varies over time, the probability of infection depends on virema in a sigmoidal fashion. The number of vectors infected is maximized when the peak viremia is just below the lethal level. These relationships are likely to vary between vector and host species, and should be considered when evaluating the relative importance of different host and vector species to transmission in arbovirus systems such as West Nile virus.