For quite some time the achievements of students in remote Australian schools have been lamented. There is not necessarily anything new about the relative difference between the results of Aboriginal and Torres Strait Islander students in remote communities and their counterparts in urban, regional and rural schools across Australia. However, in the last decade a number of changes in the education system have led to the difference being highlighted--to such an extent that what had been an "othering" of remote students (and their families) has turned into marginalisation that is described in terms of disadvantage, deficit and failure. One of the primary instruments used to reinforce this discourse has been the National Assessment Program--Literacy and Numeracy (NAPLAN) testing. This instrument has also been used as part of the justification for a policy response that sees governments attempting to close the educational gap, sometimes through punitive measures, and sometimes with incentives. At a strategic level, this is reflected in a focus on attendance, responding to the perceived disadvantage, and demanding higher standards of performance (of students, teachers and schools more generally). Accountability has resulted in lots of counting in education--counts of attendance, enrollments, dollars spent and test scores. These measures lead one to conclude that remote education is failing, that teachers need to improve their professional standards and that students need to perform better. But in the process, have we who are part of the system lost sight of the need to make education count? And if it is to count, what should it count for in remote Aboriginal and Torres Strait Islander communities? These are questions that the Cooperative Research Centre for Remote Economic Participation is attempting to find answers to as part of its Remote Education Systems project. This article questions the assumptions behind the policy responses using publicly available NAPLAN data from very remote schools. It argues that the assumptions about what works in schools generally do not work in very remote schools with high proportions of Aboriginal and Torres Strait Islander students. It therefore questions whether we in the system are counting the right things (for example attendance, enrollments and measures of disadvantage).