OBJECTIVES: To assess i) how network meta-analyses (NMAs) included within manufacturer submissions to the National Institute for Health and Care Excellence (NICE) have been criticised by its Evidence Review Groups (ERGs); ii) how some of these criticisms might be avoided in future submissions; and iii) the extent to which such avoidance might increase the likelihood of a new intervention being approved. METHODS: We reviewed the ERG reports of all NICE technology appraisals published since January 2007 to identify those where the manufacturer’s submission included an NMA. Subsequently, all criticisms made by the ERG of such analyses were analysed to seek common themes; and assess how often any one type of criticism was associated with a rejection by NICE. RESULTS: A total of 181 NICE technology appraisal reports were evaluated. These covered 243 separate interventions, 83 (34%) of which were drugs for cancer. Overall 37–64% of submissions cited NMAs, of which 43–83% were criticised, with this proportion having increased over time. Avoidable criticisms related to flaws in the systematic review methodology used to identify relevant RCTs for the analysis; inappropriate pooling of data from heterogeneous studies; and use of suboptimal statistical approaches in conducting the NMA. Unavoidable criticisms related to the lack of RCTs available for competitor drugs in the population of interest. However, no association was found between flaws in the NMA and a decision by NICE not to approve the use of the intervention.  Instead, such rejection was associated mainly with a lack of evidence of clinical efficacy or cost-effectiveness in the target population. CONCLUSIONS: Most criticisms of NMAs could be avoided by a more rigorous and transparent approach to conducting and reporting the underlying systematic review and statistical analysis. However, rejection of submissions remains a considerable risk where the underlying evidence is weak.