Abstract This paper describes a limited longitudinal study of young children's early numeracy development within three testing cycles, at the mid‐point and towards the end of their reception year (at five years‐of‐age) and again at the mid‐point of Year 1 (at six years‐of‐age), located within the broader context of progress through to Key Stage 1 SAT results (at seven years). Assessment was carried out using the Utrecht Early Mathematical Competence Test (Van Luit et al., 1994). This comprised eight sub‐topics, five items in each, including comparison, classification, correspondence, seriation, counting, calculation and real‐life number problem solving. Broadly, one set of sub‐tests related to understanding of relations in shape, size, quantity and order, whilst a second set of sub‐tests related to basic arithmetic. Three hundred pupils were selected from twenty‐one schools, large and small, from rural and urban areas, with high and low concentrations of children eligible for free school meals and/or with special educational needs, as well as representing a broad range of achievements levels based on standards assessment tasks. Whilst this paper focuses upon the performance of English pupils, reference is also made to the larger European sample which involved children from Flemish‐speaking Belgium, Germany, Greece, Slovenia and the Netherlands.Results showed that children's total scores at around the mid‐point of reception year were indeed predictive of later achievement at the end of KS1 though the combined scores over three testing cycles which extended to the mid‐point of Year 1, were more so. Discriminant analysis confirmed that a combination of a counting sub‐test (one seemed sufficient) and a sub‐test focusing on understanding of relations in shape, size, order or quantity (a different one at each testing cycle), together with the general number knowledge sub‐test was best predictive of final SAT levels.Comparison with the international data set suggested a different trajectory for English pupils, with more of a bias towards arithmetic sub‐tests than their European counterparts who start school later. Moreover, the pattern of dependence of scores on age in which no advantage was found in including any national differences was especially interesting. These findings are discussed within the context of different school start ages and traditions of preparation for formal schooling. Perhaps what emerged most strongly is the need for young English pupils to maintain a broad and balanced early mathematics curriculum, which places appropriate emphasis on practical problem solving.