Abstract Background Since the start of the COVID-19 outbreak, the Belgian government has implemented various infection prevention and control measures. This study assessed the extent to which the general population in Belgium adhered to these measures, and which determinants were associated with adherence. Methods We undertook an internet survey among a sample of the Belgian population, representative for sex, age, socio-economic status and province. The questionnaire included various demographic, socio-economic and health-related questions, and also drew upon the Protection Motivation Theory as a theoretical framework to measure levels of perceived severity, vulnerability, perceived usefulness of the measures (response efficacy), perceived personal capacity to adhere (self-efficacy), and past and future adherence. Data were collected in Dutch and French, the main languages of Belgium. Results Our study was carried out in September 2020, and the number of respondents was 2008. On average, respondents provided high scores for each of the measures in place in September in terms of response efficacy (range of 3.54–4.32 on 1 to 5 Likert-scale), self-efficacy (range of 3.00–4.00), past adherence (4.00–4.68) and future adherence (3.99–4.61). The measure that overall received the highest scores was wearing a face mask in public spaces, while ‘the social bubble of 5’ generally received the lowest scores. There was a statistically significant relationship between response efficacy and self-efficacy on the one hand and (past and future) adherence on the other hand, in a multivariate model corrected for confounders. Vulnerability and severity did not show statistical significance. Conclusion Risk communication regarding COVID-19 should place a stronger emphasis on helping people understand why implemented measures are useful and how they can be put into practice, more than on increasing fear appeals.