We discuss how current and future data on the clustering and number density of z~3 Lyman-break galaxies (LBGs) can be used to constrain their relationship to dark matter haloes. We explore a three-parameter model in which the number of LBGs per dark halo scales like a power-law in the halo mass: N(M) = (M/M_1)^S for M>M_m. Here, M_m is the minimum mass halo that can host an LBG, M_1 is a normalization parameter, associated with the mass above which haloes host more than one observed LBG, and S determines the strength of the mass dependence. We show how these three parameters are constrained by three observable properties of LBGs: the number density, the large-scale bias, and the fraction of objects in close pairs. Given these three quantities, the three unknown model parameters may be estimated analytically, allowing a full exploration of parameter space. As an example, we assume an LCDM cosmology and consider the observed properties of a recent sample of spectroscopically confirmed LBGs. We find that the favored range for our model parameters is M_m = (0.4-8)x10^10 Msun/h, M_1 = (6-10)x10^12 Msun/h, and 0.9 < S < 1.1. The preferred region in M_m expands by an order of magnitude and slightly shallower slopes are acceptable if the allowed range of bias is permitted to span all recent observational estimates. We also discuss how the observed clustering of LBGs as a function of luminosity can be used to constrain halo occupation, although due to current observational uncertainties we are unable to reach any strong conclusions. Our methods and results can be used to constrain more realistic models that aim to derive the occupation function N(M) from first principles, and offer insight into how basic physical properties affect the observed properties of LBGs.

13 pages, 5 figures, Submitted to MNRAS, minor corrections