Previous studies have found biochar-induced effects on native soil organic carbon (NSOC) decomposition, with a range of positive, negative and no priming reported. However, many uncertainties still exist for which parameters driving the amplitude and the direction of the biochar priming. Materials and methods: We conducted a quantitative analysis of 1170 groups of data from 27 incubation studies using boosted regression trees (BRT). BRT is a machine learning method combining regression trees and a boosting algorithm, which can effectively partition independent influences of various factors on the target variable in the complex ecological processes. Results and discussion: The BRT model explained a total of 72.4% of the variation in soil carbon (C) priming following biochar amendment, in which incubation conditions (36.5%) and biochar properties (33.7%) explained a larger proportion than soil properties (29.8%). The predictors that substantially accounted for the explained variation included incubation time (27.1%) and soil moisture (5.0%), biochar C/N ratio (6.2%), nitrogen content (5.5%), pyrolysis time during biochar production (5.1%), biochar pH (4.5%), soil C content (5.2%), sand (4.7%) and clay content (4.1%). In contrast, other incubation conditions (temperature, biochar dose, whether nutrient addition), biochar properties (biochar C, feedstock type, ash content, pyrolysis temperature, whether biochar being activated), and soil properties (nitrogen content, silt content, C/N ratio, pH, land use type) had small contribution (each