OBJECTIVES: Severe (requiring assistance from another individual) hypoglycaemic events (SHEs) are important from both clinical and economic perspectives. The aim was to assess treatment-related SHEs risk in different treatment regimens in type 1 and 2 diabetes. METHODS: Anti-diabetic treatments were stratified into six groups: basal-bolus (BB); basal in combination with oral therapy (BOT); insulin pumps; biphasic insulin; sulfonylureas; other than SU oral antidiabetic drugs (OADs). Insulin treatments were further split into analogues (IA) and human insulins (HI). The systematic review of Medline, EMBASE and Cochrane databases was conducted for recent (≤10 years), large (n ≥100), observational studies. Data on time horizon, participants number and SHEs occurrence were extracted. Using a random effects Poisson model within MCMC framework we estimated the SHEs rates. RESULTS: In the systematic review 5220 publications were found, 525 full texts evaluated and 101 articles included (55 trials). The following average SHE/year (No of studies; 95%CI) were estimated: Type 1: BB with IA as the basal component: 0.53 (7; 0.29–1.18); BB with HI: 1.10 (6; 0.58–2.71); pump treatment: 0.18 (14; 0.13–0.25); biphasic insulin: 1.10 (0.58–2.71). Type 2: BOT with IA as the basal component: 0.13 (11; 0.04–1.17); BOT with HI: 0.21 (7; 0.08–0.88); BB with IA: 0.01 (6; 0.003–0.25); BB with HI: 0.56 (3; 0.16–9.65); biphasic IA: 0.10 (12; 0.05–0.26); biphasic HI: 0.20 (6; 0.07–0.93); sulfonylureas: 0.05 (6; 0.02–0.14); OADs: 0.01 (0.001–0.008). The differences in the SHEs risk among regimens were significant at a 95% level. CONCLUSIONS: The SHEs risk differs in both type 1 and 2 diabetes across various treatment regimens. HI was found to increase the SHEs risk compared to IA. Limited availability of studies and heterogeneous data make it difficult to come up with precise rate estimation for typical anti-diabetic treatment regimens.