Hence the core intervention was screening and personalised feedback, with the option of a more extended intervention for those who wanted. This study found that among high-risk participants, there were significantly greater reductions in weekend LY2109761 drinking and drinking to intoxication in participants receiving the intervention compared with those receiving the control, whereas no difference in drinking outcomes was found between experimental groups in low-risk participants. The inclusion of non-drinkers or light drinkers was thought to dilute the intervention effect and offer a possible explanation for the neutral finding in a trial of online screening and personalised feedback on multiple risk behaviours, based in a student health centre. Participants in this study were aged 17–24, half were female and all were university students. At six week follow-up, there was no difference between groups intervention, 2) assessment-only and 3) minimal contact) in the proportion of students drinking within recommended limits for binge drinking. The inclusion of low-risk drinkers does not explain the neutral finding in our trial as all participants were drinking above recommended limits. There is no internationally agreed cutoff score for the AUDIT-C, with advocated thresholds for detecting hazardous drinking ranging from 2 to 5 in women and 3 to 6 in men. The AUDIT-C cut-off score of 5 or more used in this trial reflects clinical guidance in England and was most accurate in detecting drinking above UK weekly limits in a trial of people seeking help with their drinking online. The neutral findings of this trial do not appear to be explained by a floor effect as our post-hoc analysis of participants with a baseline AUDIT-C score of eight or more found no benefits for the intervention. A challenge that faces the interpretation of many trials of online interventions is that they evaluate access to the intervention, rather than engagement or use of it. We do not know whether people read the feedback, particularly when it was presented alongside feedback on other health behaviours. Alternatively, the feedback may not have been perceived as relevant or valid, for example if the recommended limits are not seen as reliable. It may also be that the type of feedback was not effective at reducing alcohol intake, where effective online SBI in student populations often includes normative feedback. It is possible that the control condition may have been contaminated as the trial was conducted in one organisation although this would have required employees to share information about their responses to the questionnaire and the feedback obtained. It is also questionable whether personalised feedback delivered to someone else would impact on another person’s drinking behaviour. This trial was supported by a small budget and conducted within a tight timeframe which militated against a qualitative exploration of the experience of people taking part in this study which may have illuminated the neutral findings. Future studies in this field would benefit from exploring the feasibility of delivering an online health check in the workplace, by considering the issues that affect participation and engagement with the intervention, along with the challenges of conducting a trial in this setting.