Existing models of the basic case reproduction number (R0) for vector-borne diseases assume (i) that the distribution of vectors over the susceptible host species is homogenous and (ii) that the biting preference for the susceptible host species rather than other potential hosts is a constant. Empirical evidence contradicts both assumptions, with important consequences for disease transmission. In this paper we develop an Ideal Free Distribution (IFD) model of host choice by blood-sucking insects, predicated on the argument that vectors must have evolved to choose the least defensive hosts in order to maximize their feeding success. From a re-analysis of existing data, we demonstrate that the interference constant, m, of the IFD can vary between host species. As a result, the predicted distribution of insects over hosts has 2 desirable and intuitively plausible behaviours: that it is heterogeneous both within and between host species; and that the intensity of heterogeneity varies with host and vector density. When the IFD model is incorporated into R0, the relationship with the vector: host ratio becomes non-linear. If correct, the IFD could add considerable realism to models which seek to predict the effect of these ecological parameters on disease transmission as they vary naturally (e.g. through seasonality in vector density or host population movement) or as a consequence of artificial manipulation (e.g. zooprophylaxis, vector control). It raises the possibility of targeting transmission hot spots with greater accuracy and concomitant reduction in control effort. The robustness of the model to simplifying assumptions is discussed.