Mastery of the cardinality principle (CP) is foundational for children's mathematical development. Children's age and subitizing and counting skills contribute to their CP acquisition. This study aimed to investigate the effectiveness of different types of instruction for supporting 3-4-year-olds' CP development and its interplay with child characteristics, following a pretest-intervention-posttest-retention-test design. The type of instruction given during the intervention was manipulated in four conditions, i.e., counting only, labeling only, first counting and next labeling, or first labeling and next counting. We offered 180 3-4-year-olds three tasks to assess their CP mastery: Hidden Stars, Give-N, and Puppet Miscounting. We also assessed their subitizing, verbal counting, and synchronous counting skills. Instruction involving first labeling and next counting was most effective to support children's CP development from pretest to posttest. This effect did not persist at the retention test. When including child characteristics in the analyses, condition no longer contributed to CP development, but counting and subitizing skills did. Moreover, our analyses pointed to a complex interplay between children's counting and subitizing skills and condition. Instruction involving both counting and labeling was most effective for children with better pretest subitizing and counting skills. The results question previous studies on instruction effectiveness that did not take into account children's subitizing and counting skills. They also emphasize the need for further research on the complex interplay between effective CP instruction and child characteristics. These and future findings provide building blocks for optimizing the timing and content of early mathematics instruction.