Standardised tasks and a repeat design were used to investigate development in narrative writing by 112 primary schoolchildren. The scripts comprised the NFER "Literacy Impact" Writing Test B, completed near the end of the children's second terms in Year 5 and Year 6. The test includes a narrative task using content of appeal to both genders. The scripts were rated by specially trained panels, using a numeric scheme applied to five constituents of writing, and text-level and technical accuracy rating scales derived from relevant sources. All constituents of writing showed an improvement, although analysis of text-level ratings showed that there were substantial proportions of children whose writing included a feature in Year 5 but not in Year 6, as well as vice versa. Qualitative analysis revealed common features within attainment subgroups in content, language use, technical accuracy and overall effectiveness of the writing. The study provides findings that may contribute to the discourse of literacy education and also raises issues of interest for further comparative studies. (Contains 5 figures and 6 tables.)