Little is known about longitudinal profiles of mathematics achievement spanning all phases of basic education of the same learners. This would be useful for identifying crucial points for intervention and informing learner subject choices. This research was aimed at establishing such a profile for boys attending a quintile 5 single gender school. Learners' promotion marks across eight cohorts from Grades 1-12 were used to compare achievement trends for those who took mathematics (n = 302) or mathematical literacy (n = 160) to Grade 12. Additionally, we explored the effectiveness of mathematics marks from lower grades for predicting achievement in subsequent grades. Data were analysed fitting a Mixed Model for Repeated Measures and using regression analysis. Findings show consistently higher achievement by the mathematics set from Grade 1 to Grade 9. Relatively stable achievement was found in the primary grades up until Grade 6, where marks dropped noticeably but peaked in Grade 7, after which marks declined dramatically to Grade 9, particularly for the mathematical literacy set. After adoption of mathematical literacy, these learners' marks improved noticeably to Grade 12 with more pronounced gains the lower the Grade 9 achievement. Those who opted for mathematics continued to achieve slightly lower marks than in Grade 9 throughout the further education and training phase, with their lowest marks attained in Grade 11. The moderate to strong correlations between most grades made it possible to predict subsequent marks based on marks achieved in earlier grades. Suggestions made from the observed trends could guide appropriate early intervention programmes and subject choice.