This paper reports on the analysis of data from approximately 30,000 Pasifika students whose teachers participated in the Numeracy Development Project (NDP) between 2002 and 2005. Most students' performance improved from the beginning of the year to the end, and performance and progress seemed to improve from 2002 to 2005. As a result, the gap between European and Pasifika students appeared to reduce fairly steadily over tune. These improvements coincided with changes in the composition of the cohort over time, most notably a reduction in the percentage of students from low-decile schools and an increase in the percentage of students from medium- and high-decile schools. Hence, it is difficult to conclude with any confidence that it is the NDP that is primarily responsible for the improvements. Although the gaps in achievement between European and Pasifika students were not completely eliminated, when these differences were put beside those found in other large-scale studies, it was evident that NDP differences were much smaller (a quarter of a standard deviation compared to a whole standard deviation). The use of an individual, orally presented assessment tool with an emphasis on explaining the strategies used to get answers, rather than a written test on which the number of correct answers is simply totalled, may help to explain the positive outcomes for NDP students.