Abstract: We present an online tool – “Vaccinpraat” – that monitors messages expressing skepticism towards COVID-19 vaccination on Dutch-language Twitter and Facebook. The tool provides live updates, statistics and qualitative insights into opinions about vaccines and arguments used to justify antivaccination opinions. An annotation task was set up to create training data for a model that determines the vaccine stance of a message and another model that detects arguments for antivaccination opinions. For the binary vaccine skepticism detection task (vaccine-skeptic vs. non-skeptic), our model obtained F1-scores of 0.77 and 0.69 for Twitter and Facebook, respectively. Experiments on argument detection showed that this multilabel task is more challenging than stance classification, with F1-scores ranging from 0.23 to 0.68 depending on the argument class, suggesting that more research in this area is needed. Additionally, we process the content of messages related to vaccines by applying named entity recognition, fine-grained emotion analysis, and author profiling techniques. Users of the tool can consult monthly reports in PDF format and request data with model predictions. The tool is available at https://vaccinpraat.uantwerpen.be/