The study set out to explore different mathematical difficulties among 877 second-grade children and to test the effect of memorization versus conceptual practices with number combinations. It used a latent profile analysis of baseline measurements of digit writing speed, number combination fluency, multidigit calculation, and number sense skills to identify six latent classes: three mathematical learning difficulty classes, two typical classes, and one high-achieving class. The memorization practice produced superior improvement for all classes of students except the high-achievers class. These results suggest that memorization practice with basic number combinations should not be considered poor teaching practice. It is important for teachers to incorporate practices with number combinations that focus on speed and memorization, even for children who struggle with mathematics.