Background and aims: Lower-risk recommendations for avoiding gambling harm have been developed as a primary prevention measure, using self-reported prevalence survey data. The aim of this study was to conduct similar analyses using gambling company player data. Methods: The sample (N 5 35,753) were Norsk Tipping website customers. Gambling indicators were frequency, expenditure, duration, number of gambling formats and wager. Harm indicators (financial. social, emotional, harms in two or more areas) were derived from the GamTest self-assessment instrument. Receiver operating characteristics (ROC) curves were performed separately for each of the five gambling indicators for each of the four harm indicators. Results: ROC areas under the curve were between 0.55 and 0.68. Suggested monthly lower-risk limits were less than 8.7 days, expenditure less than 54 V, duration less than 72-83 min, number of gambling formats less than 3 and wager less than 118-140V. Most risk curves showed a rather stable harm level up to a certain point, from which the increase in harm was fairly linear. Discussion: The suggested lower-risk limits in the present study are higher than limits based on prevalence studies. There was a significant number of gamblers (5-10%) experiencing harm at gambling levels well below the suggested cut-offs and the risk increase at certain consumption levels. Conclusions: Risk of harm occurs at all levels of gambling involvement within the specific gambling commercial environment assessed in an increasingly available gambling market where most people gamble in multiple commercial environments, minimizing harm is important for all customers.  Kulcsszavak: gambling, low-risk limits, behavioural data, gamtest