Given the importance of soil organic carbon (SOC) stocks and their dynamics in the regulation of climate change, understanding the mechanisms of SOC protection from decomposition is crucial. It is recognized that soil aggregates can provide effective protection of organic carbon from microbial decomposition. Currently, there is no systematic method for estimating the amount of protected carbon within aggregates. However, differences between CO2 emissions from incubation of intact versus crushed aggregates have been widely used as a proxy for SOC physical protection within aggregates. There is no global analysis on this type of experiment yet, nor on the drivers of the amount of SOC physically protected in soils. Using a meta-analysis including 165 pairs of observations from 22 studies encompassing a variety of ecosystems, climate and soil types, we investigated the crushing effects on cumulative carbon mineralization from laboratory incubation experiments. The aggregates were initially separated by either wet sieving or dry sieving before dry crushing. Our results indicated that aggregate crushing led on average to +31 % stimulation of carbon mineralization compared with intact aggregates, which represented 0.65 to 1.01 % of total SOC. This result suggests the mineralization of a previously protected pool of labile organic carbon. The linear regression analysis showed that the crushing effect on carbon mineralization depended on soil characteristics (carbon content, clay content and pH) as well as on aggregate size. Crushing aggregates stimulated carbon mineralization relative to control, up to +63 % in large aggregates (>10 mm), +38 % in large macro-aggregates (2–8 mm), +14 % in small macro-aggregates (0.25–2 mm) and +54 % in micro-aggregates (<0.25 mm). Within each aggregate size-class, the crushing effect depended on the crushing intensity. The destruction of aggregates to <0.05 mm size had a greater effect on carbon mineralization (+130–133 %) than the destruction of aggregates to >2 mm (+3 to 40 %), < 2 mm (+58 to 62 %) and < 0.25 mm (+32 to 62 %) sizes regardless of the initial aggregate size. These results suggest that macroaggregates (>0.25 mm) are less protective than microaggregates (<0.25 mm). Our dataset also show that soil physicochemical characteristics and experimental conditions influenced more the amount of protected SOC than land use and management. Contrary to our expectations the crushing effect was not affected by tillage practices nor land use. Standardizing the experimental conditions of aggregate crushing and subsequent incubation is needed to assess and compare the amount of physically protected SOC in diverse soils, and then to better understand the processes and drivers of SOC protection inside aggregates.