OBJECTIVES:  Recent studies have quantified the resource implications associated with severe hypoglycaemia episodes (SHE) and non-severe hypoglycaemia episodes (NSHE). The objective of this study was to characterize the relationship between the frequency of SHE/NSHE and patient/therapy characteristics in insulin treated type-2 diabetes mellitus (T2DM) patients. METHODS:  We conducted a literature review of the MEDLINE database for insulin-based T2DM clinical trials where sulfonylurea usage was also reported between 1st June 2007 and 1st June 2012. Patient demographics, treatments effects and treatment type were extracted and the SHE/NSHE rate per 100 patient years (PY) was modeled via a log-linear regression model using R.  RESULTS:  Data were extracted from 82 studies for a total of 155 trial arms meeting the search criteria; treatment regimes were categorized into basal analog (30%); bolus analog (4%); basal bolus (24%); biphasic analog (25%); biphasic human (8%) and NPH (10%). Mean (standard deviation) age was 57.9 (3.8) years; 10.2 (2.4) years duration of diabetes; 8.8% (0.7)baseline and 7.5% (0.5) end-of-study HbA1c respectively. Mean reported NSHE/100 PYs was 929.8. The final regression model was NSHE/100PY=exp(14.771 - 0.088 × age - 0.667 × baseline HbA1c + 0.427 × HbA1c reduction + 0.189 × duration diabetes + 0.007 × % study allowed sulphonylurea – 0.545 × basal analog insulin) with an R2 goodness of fit of 0.226.  Mean reported SHE/100 PYs was 5.9. The final regression model was SHE/100PY=exp(10.794 - 0.101 × age - 0.723 × baseline HbA1c + 0.638 × HbA1c reduction + 0.163 × duration diabetes + 0.768 × biphasic insulin )-1 with an R2 of 0.219. CONCLUSIONS: This study provides a basis for predicting the number of SHE and NSHE in insulin treated T2DM subjects.  When linked to resource and quality life, these equations may facilitate the improved estimation of the health economic burden associated with SHEs and NSHEs.