Cisplatin (ddp), which is commonly employed in the treatment of many advanced cancers, often results in initial therapeutic success; however, rapid progression of ddp-resistant cells remains the main reason for treatment failure. Facd with such a problem, we investigated the fitness differences between ddp-sensitive and ddp-resistant cell lines. We found that the growth of ddp-resistant cells was significantly slower than that of sensitive cells due to elevated ROS levels, which suggested that the ddp resistance mechanisms may have negative impacts on the growth of resistant cells. Furthermore, we observed that, when mixed with ddp-sensitive cells, ddp-resistant cells failed to compete, and the growth of ddp-resistant cells could therefore be suppressed by treatment in vivo. We propose a mathematical model parameterized based on in vivo experiments to describe the allometric growth of tumors consisting of two competing subclones. According to our model, a quantitative strategy with a variant drug-dosing interval is proposed to control tumor growth. Taking advantage of intratumoral competition, our strategy with appropriate dosing intervals could remarkably delay the development of ddp resistance and prolong overall survival. Maintaining a certain number of ddp-sensitive cells rather than eradicating the tumor with continuous treatment is feasible for future tumor treatment. © 2017 The Author(s).