Computer-assisted interventions designed to remediate low numeracy and developmental dyscalculia (mathematical learning disability) have been utilised in preschools and kindergartens with some efficacy for over thirty years (Clements, 2002). A recent development in this field is ‘adaptive game’ technology, which adapts task difficulty online as children learn. The Number Race is the first such package for mathematics. Previous efficacy studies suggest its use results in an improvement in core measures of early numeracy, such as speed at enumerating 1-3 objects (subitizing) and comparison of numerals and groups of objects. The present study tested the efficacy of a new version of The Number Race (version 3.0) using New Zealand English and incorporating new instructional factors, in a younger population than most previously tested. Participants were twelve 5-to-7-year-old children and a typically developing control group matched on age and sex (n = 12). Following pre-testing using standardised tests and a computerised battery, children in the intervention group used The Number Race for twenty minutes each school night, for one month. Post-testing results showed that there was a significant improvement in counting and subitizing speed for the intervention group. Participants also became faster and more accurate at comparing numerals. There were no significant changes in standardised mathematics scores. The mental number line task did not show any significant differences before and after intervention but a wide variety of patterns and possible use of strategies were revealed. Overall, this new version of The Number Race seems to have modest effects in this population.