Intercropping has been advocated as an environmentally benign method to suppress weeds in agriculture. However, it is not evident from the literature what size of weed suppressive effect is achieved on average by intercropping, and how species choice and crop management affect this effect. We conducted a global meta-analysis of published data to quantify the effect of intercropping on weed biomass in annual arable intercrops grown for their final product. We searched the literature to identify all papers reporting usable experimental data and extracted 339 data records from 39 publications containing data from 76 independent experiments. Two metrics of weed suppression were defined to assess the weed suppressive effect of intercropping: the ratio of observed weed biomass in an intercrop to weed biomass in the less weed suppressive sole crop (Rweak), and the ratio of weed biomass in the intercrop to weed biomass in the more weed suppressive sole crop (Rstrong). On average, weed biomass in the intercrop was substantially and significantly (58%) lower (Rweak = 0.42) than in the less suppressive sole crop. No significant difference was found between weed biomass in the intercrop and weed biomass in the more weed suppressive sole crop, even though weed biomass tended to be slightly larger in the intercrop than in the more weed suppressive sole crop (Rstrong = 1.08). Findings were consistent across different groups of species combinations, such as maize/legume and small-grain cereal/legume intercrops. Intercrops with an additive design had stronger weed suppression than intercrops with a replacement design. In the latter, a mixed arrangement gave stronger weed suppression than a row design, while spatial arrangement did not affect weed suppressive ability in additive designs. No significant effects on weed biomass were found of simultaneous vs. relay intercropping, and of nitrogen fertilizer input. The Rweak decreased significantly with the land equivalent ratio in additi