Cardiovascular disease (CVD) remains a pressing global health issue that poses substantial threats to public well-being. The modified cardiometabolic index (MCMI), a newly developed composite indicator linked to insulin resistance (IR) and central obesity, has not been extensively explored in terms of its association with the occurrence of CVD. To fill this research gap, we carried out a large-scale retrospective cohort study to examine how MCMI is associated with the onset of CVD. Additionally, we compared MCMI with the triglyceride-glucose (TyG) index to evaluate the ability of both indices to predict CVD risk. This retrospective cohort study drew data from the China Health and Retirement Longitudinal Study (CHARLS). It included participants who were 45 years of age or older, with baseline assessments conducted in 2011 and follow-up evaluations in 2020. The diagnosis of CVD was based on self-report. To explore the association between MCMI and the incidence of CVD, we used logistic regression models, restricted cubic splines, and subgroup analyses. Moreover, receiver operating characteristic (ROC) analysis was performed to assess the predictive performance of both MCMI and the TyG index for CVD occurrence. Among the 6,117 participants in the total population, 1,298 eventually developed CVD. MCMI was identified as an independent risk factor for CVD development, with an adjusted odds ratio of 1.18 [95% confidence interval (CI): 1.08-1.29]. A J-shaped non-linear association between MCMI and the incidence of CVD was also discovered. Subgroup analyses showed that MCMI could stably predict CVD. In the comparison of predictive abilities, MCMI demonstrated a statistically superior but still modest predictive power compared to the TyG Index, as evidenced by area under the curve (AUC) values of 0.581 for MCMI and 0.559 for the TyG index (p < 0.001). This study indicated that the MCMI may serve as a preliminary screening indicator for cardiovascular disease risk in middle-aged and elderly populations. Although its independent predictive performance is limited, its simplicity and cost-effectiveness render it potentially useful in primary care settings and large-scale screening initiatives.