Dietary fibre is believed to provide important health benefits including protection from colorectal cancer. However, the evidence on the relationships with different dietary fibre sources is mixed and little is known about which fibre source provides the greatest benefits. We conducted a dose–response meta-analysis of prospective cohorts to summarise the relationships of different fibre sources with colorectal cancer and adenoma risks. Analyses were restricted to publications that reported all fibre sources (cereals, vegetables, fruits, legumes) to increase comparability between results. PubMed and Embase were searched through August 2018 to identify relevant studies. The summary relative risks (RR) and 95 % CI were estimated using a random-effects model. This analysis included a total of ten prospective studies. The summary RR of colorectal cancer associated with each 10 g/d increase in fibre intake were 0·91 (95 % CI 0·82, 1·00; I2 = 0 %) for cereal fibre, 0·95 (95 % CI 0·87, 1·03, I2 = 0 %) for vegetable fibre, 0·91 (95 % CI 0·78, 1·06, I2 = 43 %) for fruit fibre and 0·84 (95 % CI 0·63, 1·13, I2 = 45 %) for legume fibre. For cereal fibre, the association with colorectal cancer risk remained statistically significant after adjustment for folate intake (RR 0·89, 95 % CI 0·80, 0·99, I2 = 2 %). For vegetable and fruit fibres, the dose–response curve suggested evidence of non-linearity. All fibre sources were inversely associated with incident adenoma (per 10 g/d increase: RR 0·81 (95 % CI 0·54, 1·21) cereals, 0·84 (95 % CI 0·71, 0·98) for vegetables, 0·78 (95 % CI 0·65, 0·93) for fruits) but not associated with recurrent adenoma. Our data suggest that, although all fibre sources may provide some benefits, the evidence for colorectal cancer prevention is strongest for fibre from cereals/grains.