This study revisits the nexus between debt and growth in the following distinct ways. First, the interactive fixed effects and dynamic panel threshold methods are employed to estimate the threshold effect of public debt on economic growth. Secondly, we account for cross-section dependence, time-varying unobserved heterogeneity, and feedback effects. Thirdly, we provide a more complete picture of the debt-growth nexus at different income levels by decomposing our sample into upper-middle-income and low-income subsamples. The empirical application based on a sample of 62 emerging and developing countries from 2000 to 2018 suggests that inverted U-shaped nexus exists between debt and growth in emerging and developing countries. However, the dynamic panel threshold results indicate that public debt harms growth when the indebtedness level exceeds the estimated threshold of 50.19% and 25.09% of GDP for the upper-middle-income and low-income subsamples, respectively. Furthermore, the negative effect of debt seems to be stronger in low than in upper-middle-income countries. By comparison, evidence of a positive effect of public debt on economic growth is weak and insignificant in low-income countries.