PURPOSE: It is often difficult or impossible to use network meta-analysis (NMA) to compare the treatment effects of competing interventions using randomized controlled trial (RCT) evidence alone. Our purpose is to describe methods to supplement RCT evidence using single-arm studies, non-randomized trials, or observational evidence and clarify when they are appropriate. DESCRIPTION: NMA uses multiple RCTs to simultaneously estimate many relative treatment effects and models exist for various outcome types; it is thus our preferred framework. However, RCT evidence for interventions of interest can be disconnected or provide only highly uncertain estimates for relative effects. Several alternative methods of indirect comparison and extensions of NMA itself have been proposed to supplement the RCT evidence. When only aggregate data (AD) is available, NMA can be extended to 3-level hierarchical models, with the additional level representing observational or RCT effects. Random effects could also be placed on the study baseline to include single arm studies in the NMA framework. When individual patient data (IPD) is available from at least one RCT alternative indirect comparison methods have been proposed, such as propensity score reweighting (matching adjusted indirect comparison) and outcome-regression models (simulated treatment comparison). We will discuss these methods in the context of specific clinical applications from our own research and from the literature, and explain where they produce biased estimates or make inappropriate assumptions.