In order to help students to spell accurately, both students and teachers need to be aware of the causes of their spelling errors. Several methods have been developed for analysing spelling errors. Some researchers, such as Bruck and Waters (1988) and Bebout (1985), have categorised the phonetic errors that were made. Others, such as Brooks et al (1993) and Elliott and Johnson (2008) have included particular spelling rules and letter patterns too. Elliott and Johnson's (ibid) categories related the errors to spelling conventions. They identified misspelt words in random stratified samples of single sentences taken from 16-year-old students' narrative writing in an English examination, and grouped them into five categories and 22 sub-categories of spelling errors. The precision of their categories, along with the similarity of their sample, meant that a modified version of their method was considered the most appropriate for this study. The first objective of this study was to develop a method for transforming their categories into a method that could be used to categorise spelling errors in larger samples of writing. The second objective was to use the new method to code samples from different years to see whether the proportion of errors attributed to each category by Elliott and Johnson (ibid) represented the proportions seen in a larger sample of writing. This would show whether their framework could be applied to different samples, for example English as a second language students, or students of different ages. This paper details the method and expected outcomes of the study.