Background: GRADEpro software was developed to provide a transparent and logical approach to grading evidence and strength of recommendations or implications for practice for interventional studies. GRADEpro uses a default for rating the quality of evidence based on the study design (RCT or observational) and this is downgraded or upgraded based on other factors including limitations, inconsistency, indirectness and imprecision.

Objectives: To examine the challenges of using GRADEpro to different types of reviews and present some possible solutions to the problems. We used GRADEpro to assess the quality of evidence for two NICE clinical guidelines.

Methods: Three worked examples illustrating the problems of using GRADEpro in different types of reviews will be presented. These are: i) assessment of non-comparative data using GRADEpro, ii) rating imprecision where the clinical cut-offs for potential benefits and harms are unclear, and iii)presentation of findings where denominators differ between studies.

Results: Some practical solutions to commonly encountered reviewing problems will be presented and potential for application of the GRADE approach across a range of study types discussed.

Conclusions: Although GRADE is a methodologically rigorous method that has been well developed for use in intervention studies questions remain about its consistency and objectivity, especially when applied to different study types. Further development will help to improve the GRADE approach further but this needs to be undertaken with extensive field testing as this is the only way to ensure its methods stand up across a wide range of situations.