An inverse association between coffee consumption and the risk of colorectal cancer has been found in several case-control studies, but such an association was not consistent in prospective cohort studies. We conducted a systematic meta-analysis of prospective cohort studies on coffee consumption and colorectal cancer published up to June 2008. We combined relative risks (RR) for colorectal cancer comparing high vs. low categories of coffee consumption using random-effects models. We identified 12 eligible cohort studies, which included 646,848 participants and 5,403 cases for colorectal cancer. The summarized result of the meta-analysis comparing high- vs. low-consumption categories showed no significant effect of coffee consumption on colorectal cancer risk (RR = 0.91; 95% confidence intervals [CI]: 0.81-1.02). The RR was 0.93 (95% CI: 0.71-1.22) when considering 4 studies conducted in the United States of America, 0.91 (95% CI: 0.76-1.10) for 5 studies from Europe, and 0.83 (95% CI: 0.62-1.10) for 3 Japanese studies. No significant differences by sex and cancer-site were found, but there was a slight suggestion of an inverse association between coffee consumption and colon cancer in women (RR = 0.79; 95% CI: 0.60-1.04), especially Japanese women (RR = 0.62; 95% CI: 0.37-1.05). The suggestive inverse associations were slightly stronger in studies that controlled for smoking and alcohol, and in studies with shorter follow-up times. Information on coffee type, its serving size, or brewing method may provide a better understanding of this reassuring result and the real role of coffee on colorectal cancer risk.