Urbanization and industrialization have significant impacts on energy consumption and CO2 emissions, but their relationship varies at different stages of economic development. Taking cognizance of heterogeneity and the “ratchet effect,” this paper adopts the Stochastic Impacts by Regression on Population, Affluence, and Technology (STIRPAT) framework as a starting point and re-estimates the relationship using different panel date models. The main results are obtained by dynamic panel threshold regression models, which divide a balanced panel dataset of 73 countries over the period of 1971–2010 into four groups according to their annual income levels. The key results are: (1) in the low-income group, urbanization decreases energy consumption but increases CO2 emissions; (2) in the middle-/low-income and high-income groups, industrialization decreases energy consumption but increases CO2 emissions, while urbanization significantly increases both energy consumption and CO2 emissions; (3) for the middle-/high-income group, urbanization does not significantly affect energy consumption, but does hinder the growth of emissions; while industrialization was found to have an insignificant impact on energy consumption and CO2 emissions; (4) from the population perspective, it produces positive effects on energy consumption, and also increases emissions except for the high-income group. These novel methodology and findings reveal that different development strategies of urbanization and industrialization should be pursued depending on the levels of income in a bid to conserve energy and reduce emissions.