There is speculation on an association between sweetened, carbonated beverage consumption and cancer risk. This study aimed to examine this issue. Over 50 independent estimates of risk were available, 11 for colas specifically. A random-effects meta-analysis was carried out with tests for publication bias performed as well as Higgins and Thompson's I measure of the percentage of heterogeneity between studies that could not be explained by chance. Over all the different sites of cancer, the summary relative risk (SRR), when all 55 independent estimates were considered together, was SRR=1.03 [95% confidence interval (0.96; 1.11)]. When individual cancer sites were considered, there was no significant increase or decrease in the meta-analysis estimate of risk of cancer of the pancreas, bladder, kidney, squamous cell or adenocarcinoma of the oesophagus, colon, gastric cardia, gastric noncardia, prostate, breast, larynx and ovary or of the oral cavity, pharynx or glioma. There was no evidence in a sensitivity analysis from those studies that reported results separately for colas of an associated risk of pancreas cancer [SRR=1.00, 95% confidence interval (0.61; 1.65)]. The results for all other forms of cancers were considerably hampered by poor methodology and small numbers of studies (mainly one report on each cancer site studied). Overall, the findings are reassuring in terms of the association between soft drinks, including colas, and cancer risk, although the quality of many of the studies is quite poor by acceptable, modern standards and no study has been carried out with use of carbonated beverages as a primary hypothesis.