Children who struggle to learn maths during the early school years face life-long challenges, including grade repetition and vocational disadvantage. In the most recent Trends in International Mathematics and Science Study (TIMSS, 2011), 30% of Australian Grade 4 students performed below the minimum international proficient standard in maths, and this increased to 37% of Grade 8 students. New interactive technology in cognitive training has been shown to improve a nominated target skill by continually adapting the difficulty level of the training task to the child’s own performance. Such a program targeting early maths skills could be implemented with large groups of children. Whether this approach can improve children’s maths performance in the classroom has not been evaluated.This pilot comparative study aimed to examine (i) the feasibility of teachers delivering to whole classes an adaptive computerised maths training program developed by Cognition Matters, Sweden, and (ii) the feasibility of conducting a large trial of the program compared with usual teaching in Australian classrooms for improving maths performance.Participants were 62 students (78% recruitment rate) and their teachers from two composite primary school classes in a rural South Australian school and two Grade 1 classes in a metropolitan Victorian school. Teachers and students of the rural classes (n=24) and one metropolitan class (n=23) participated in the computerised maths training program, while the other metropolitan class continued with usual teaching (n=15). Program delivery involved teachers supervising their whole class performing daily individual training sessions on iPads over 7-8 weeks. In the training group 57% completed ≥75% of the sessions, for whom IQ scores ranged from 62 (lower-extreme) to 128 (above average) and baseline maths test scores ranged from 59 (very-low) to 137 (superior). Follow-up testing was completed by 85% of the sample; the training group showed mainly moderate improvements on the two maths tests (d=0.60, d=0.16) and two maths fluency tests administered (d=0.54, d=0.49), while the usual teaching group showed smaller improvements on the two maths tests (d=0.45, d=0.09) and one of the maths fluency tests (d=0.16).In conclusion, it is feasible for teachers in Australia to deliver a computerised maths training program to whole classes of junior primary students with different levels of IQ and maths performance. Participation rates suggest it would be feasible to conduct a larger trial of the program compared with usual teaching, which is essential to evaluate its effectiveness.