Abstract. The impact of geohazards on the mental health of local populations is well recognised but understudied. We used natural language processing (NLP) of Twitter (now known as X) posts (n=10 341) to analyse the sentiments expressed in relation to a pre-eruptive seismic unrest and a subsequent volcanic eruption in Iceland in 2019–2021. We show that although these geohazards were of a small size and caused negligible material damage, they were associated with a measurable change in expressed emotions in the local population. The seismic unrest was associated with predominantly negative sentiments (positive-to-negative sentiment ratio of 1:1.3), but the eruption was associated with predominantly positive sentiments (positive-to-negative sentiment ratio of 1.4:1). We demonstrate a cost-effective tool for gauging public discourse that could be used in risk management.