Cue-taking is thought to be influential because legislators seek information from like-minded, trusted policy experts. Unfortunately for researchers, this self-selection process complicates efforts to separate the causal effects of cues from the tendency of legislators to communicate with similar peers. Prior causally-oriented research has estimated cues' effects in exogenous networks, but not in the naturally-occurring communication networks that legislators themselves choose to form. This study examines cue-taking with two legislative field experiments, with over 2,000 observations in total, that model the diffusion of a randomly-assigned information treatment across an endogenous legislative network. Experimental results reinforce findings from classic interview-based studies of self-selected communication networks by Matthews and Stimson (1975) and Kingdon (1973): cue-taking influences a large percentage of policy positions and occurs late in the policymaking process. I also contribute to the literature by showing that on average cues complement, rather than substitute for, policy information from other sources of expertise within the legislature.