Clustered randomized trials (CRTs) are popular in the social sciences to evaluate the efficacy of a new policy or program by randomly assigning one set of clusters to the new policy and the other to the usual policy. Often, many individuals within a cluster fail to take advantage of the new policy, resulting in noncompliance behaviors. Also, individuals within a cluster may influence each other, for instance when one individual's enrollment in a new program impacts the outcomes of others due to spillover/peer effects from the new program and thus, interference is unavoidable. Current CRTs assume away noncompliance and interference, which can lead to invalid inference if any of them are present. The talk will present results on identification for CRTs when both noncompliance and interference are present. The talk also discusses estimation and inference under different forms of interference. This is joint work with Luke Keele.