The online environment has few boundaries and is evolving quickly, which may enable a greater prevalence of negative behaviour, like cyberbullying. Most cyberbullying researchers have used self-report methodologies, focussed on young people, and have had limited ability to explore contextual factors such as whether messages were retaliatory. Using stratified sampling, we selected a total of 40 YouTube® clips showing singing or dancing performances by young girls and boys. We conducted an exploratory content analysis using the first 20 comments for each of these clips (n = 800), coding eight quantitative factors and identifying themes for each comment. The key measure was the degree of positivity of each comment towards the performer in the clip, and if applicable, the degree of positivity towards the other commenters. Negative comments were directed at performers in 20% of the observed cases, whereas 74% of the comments directed at other commenters were negative, suggesting that commenters are most at risk of receiving negative comments online. Male performers received more negative comments than female. About a third of commenters appeared to criticise users for posting negative comments towards others, which is of continued interest because bystander intervention may mitigate some of the harmful effects of cyberbullying.