To compare the effect and safety of glucagon-like peptide-1 receptor agonists (GLP-1 RA) with insulin therapy on type 2 diabetes mellitus (T2DM) patients inadequately controlled with metformin and/or sulfonylurea. A systematic literature search on MEDLINE, Embase and Cochrane for randomized controlled trials (RCTs) was conducted using specific search terms ‘GLP-1 insulin type 2 diabetes clinical trials' and eight eligible studies were retrieved. Data on mean change in haemoglobin A1c (HbA1C), weight loss, fasting plasma glucose (FPG), incidence of hypoglycaemia and gastrointestinal adverse events were extracted from each study and pooled in meta-analysis. Data on postprandial plasma glucose (PPG) and adverse events were also described or tabulated. Data from eight RCTs enrolling 2782 patients were pooled using a random-effects model. The mean net change [95% confidence interval (CIs)] for HbA1c, weight loss and FPG for patients treated with GLP-1 RA as compared with insulin was −0.14% (−2 mmol/mol) [95% CI; (−0.27, −0.02)%; p = 0.03]; −4.40 kg [95% CI; (−5.23, −3.56) kg; p < 0.01] and 1.18 mmol/l [95% CI; (0.43, 1.93) mmol/l; p < 0.01], respectively, with negative values favouring GLP-1 and positive values favouring insulin. The GLP-1 group was associated with a greater reduction in PPG than the insulin group. Overall, hypoglycaemia was reported less in the GLP-1 group [Mantel–Haenszel odds ratio (M-H OR) 0.45 (0.27, 0.76); p < 0.01], while there was no significant difference in occurrence of severe hypoglycaemia [M-H OR 0.65 (0.29,1.45); p = 0.29]. A significantly higher number of gastrointestinal adverse events were reported with GLP-1 group [M-H OR 15.00 (5.44,41.35) p < 0.01]. GLP-1 RA are promising new agents compared with insulin. Further prospective clinical trials are expected to fully evaluate the long-term effectiveness and safety of these therapies within the T2DM treatment paradigm.