Soils of grasslands represent a large potential reservoir for storing CO2, but this potential likely depends on how grasslands are managed for large mammal grazing. Previous studies found both strong positive and negative grazing effects on soil organic carbon (SOC) but explanations for this variation are poorly developed. Expanding on previous reviews, we performed a multifactorial meta-analysis of grazer effects on SOC density on 47 independent experimental contrasts from 17 studies. We explicitly tested hypotheses that grazer effects would shift from negative to positive with decreasing precipitation, increasing fineness of soil texture, transition from dominant grass species with C3 to C4 photosynthesis, and decreasing grazing intensity, after controlling for study duration and sampling depth. The six variables of soil texture, precipitation, grass type, grazing intensity, study duration, and sampling depth explained 85% of a large variation (±150 g m−2 yr−1) in grazing effects, and the best model included significant interactions between precipitation and soil texture (P = 0.002), grass type, and grazing intensity (P = 0.012), and study duration and soil sampling depth (P = 0.020). Specifically, an increase in mean annual precipitation of 600 mm resulted in a 24% decrease in grazer effect size on finer textured soils, while on sandy soils the same increase in precipitation produced a 22% increase in grazer effect on SOC. Increasing grazing intensity increased SOC by 6–7% on C4-dominated and C4–C3 mixed grasslands, but decreased SOC by an average 18% in C3-dominated grasslands. We discovered these patterns despite a lack of studies in natural, wildlife-dominated ecosystems, and tropical grasslands. Our results, which suggest a future focus on why C3 vs. C4-dominated grasslands differ so strongly in their response of SOC to grazing, show that grazer effects on SOC are highly context-specific and imply that grazers in different regions might be managed differently to help mitigate greenhouse gas emissions.