Maize weevil (Sitophilus zeamais Motschulsky) is an important pest of maize (Zea mays L.) in the tropics, causing serious losses for many resource‐poor farmers who store grain on‐farm for use as food and seed. This study evaluated whether weevil resistance of six maize populations could be divergently changed by S1 and S2 selection, and assessed the importance of replicating grain samples when screening for resistance. Weevil resistance was evaluated for unreplicated S1 and for replicated and unreplicated S2 lines by infesting 50‐g grain samples with 32 young adult weevils, and then incubating the samples in a controlled temperature and relative humidity (CTH) laboratory. Divergent synthetics were formed by recombining the most resistant 10% and the most susceptible 10% of at least 100 lines screened for weevil resistance for each maize population. Replicated S2 selection was successful for both populations where it was applied, resulting in an average of 16% (P < 0.01), 49% (P < 0.05), and 20% (P < 0.01) difference between divergent synthetics for weevil progeny emerged, grain weight loss, and the Dobie index of susceptibility, respectively. S1 unreplicated selection was successful for two of the six populations, while S2 unreplicated selection was not successful. Reciprocal effects were significant (P < 0.01) for weevil resistance of F1 varietal crosses among the divergently selected synthetics, indicating the influence of maternal effects. Nevertheless, the most resistant crosses were those among the most resistant synthetics, confirming that additive gene action for weevil resistance was important. Our results provide practical insights regarding methodologies and demonstrate that it is possible to improve weevil resistance of maize populations.