Although research findings have been equivocal as to whether the use of social networking sites (SNS) increases experiences of online risk among children, the affordances of SNS lend support to this possibility, attracting much policy and public concern. The present article examines whether the use of such services increases the risks that children and young people encounter by analyzing data from a random stratified sample of approximately 1000 internet-using children aged 9-16 years in each of 25 European countries. Four hypotheses were formulated and tested. The first hypothesis, namely that children who use social networking sites will encounter more risks online than those who do not, is supported by the data. The second hypothesis stated that SNS users with more digital competence will encounter more online risk than those with less competence; this was also supported, despite being counter to common assumptions. Thirdly, we hypothesized that SNS users with more risky SNS practices (e.g. a public profile, displaying identifying information, with a very large number of contacts) will encounter more online risk than those with fewer risky practices: this too was supported by the data; thus what matters for risk is how SNS are used, a useful point for awareness-raising initiatives. The fourth hypothesis stated that SNS users with more digital competence in using the internet will experience less harm associated with online risk. The data did not support this hypothesis, since digital competence did not reduce the probability of children saying that they have been bothered or upset by something on the internet. Finally, the study found that, although this had not been predicted, whether or not risks are experienced as harmful depends on the specific relation between risks and platforms (website, instant messaging, gaming or social networking). We call on future research to explore how particular affordances sustain particular communicative conditions and, in turn, are responded to differently by children. The research and policy implications of the findings are discussed.