Life-threatening diarrhoea is observed in up to 25% of cancer patients receiving irinotecan. The associations between the UGT1A1*28 polymorphism and irinotecan-induced diarrhoea remains controversial because of conflicting data in the literature. Meta-analyses were performed on published data in terms of relationships between UGT1A1*28 and severe diarrhoea. We searched databases for relevant studies that were published in English or Chinese. Two reviewers extracted data and assessed methodological quality. UGT1A1*28 related odds ratios (ORs) were pooled by use of a fixed-effects model. The studies included were stratified into subgroups representing different races and irinotecan doses, and meta-regression analyses were performed to investigate the effect of study characteristics on the association between UGT1A1*28 and diarrhoea. Twenty trials including a total of 1760 cancer patients were included. The risk of severe diarrhoea at medium and high irinotecan doses was higher among patients with a UGT1A1*28/*28 genotype than among those with a UGT1A1*1/*1 genotype (OR=3.69, 95% confidence interval [CI]=2.00-6.83; P<0.001). Considering the patients with a UGT1A1*1/*28 genotype, the risk of toxicity was also higher than among those with a wild-type genotype at medium and high doses (OR=1.92, 95% CI=1.31-2.82; P=0.001). No association was observed between UGT1A1*28 and severe diarrhoea at low doses (<125 mg/m(2)). In conclusion, patients carrying UGT1A1*28 allele(s) are at an increased risk of irinotecan-induced severe diarrhoea. This increased risk is only apparent in those who are administrated with medium or high irinotecan doses.