Parallel versions of a new multiple-choice word-recognition test were administered to 1019 and 590 Year-1 pupils respectively. The test format was based on a published test of Word Recognition and Phonic Skills and was intended to provide reliable diagnostic information. It was found that internally consistent measures of three types of word recognition error could be derived (relating to consonants, vowels and phoneme omission), but that scales for letter order errors were unreliable. Information about the length of words which an individual can usually recognise, the proportion of high frequency to low frequency words recognised and the relative ease of recognising regular and irregular words were also considered as having possible implications for teaching. Of these measures, only word length was found to be a strong independent predictor of word recognition. Implications of these findings for the future development of norm-referenced tests that at the same time provide objective feedback about individual strengths and weaknesses are discussed.