We present work on the extension of network-based models of HIV prevention to accommodate modeling network-informed recruitment and randomization strategies in clinical trials. The underlying contact network induces dependence among the outcomes of members of a population—strongest for those who are separated by few steps in the network. In some experimental settings, cluster randomization is used to account for this dependence and create randomized units for which the assumption of independence is more plausible. When clusters are densely connected subnetworks of a local contact network, however, we cannot easily create a sampling frame of clusters to randomize. Network-based recruitment and randomization is an alternative, in which we assume people connected enough to recruit each other are member of the same subnetwork. The tools presented here make it easier to compare the benefits of individual- and cluster-randomized approaches, as well as model innovative network-informed trial designs.