Online social networks have created a platform for the human to share information at an unprecedented scale. However, misinformation on various topics also gets spread across the social media. Misinformation on vaccination especially anti-vaccination campaigns could adversely affect public health by preventing many people from taking the vaccination, resulting in poor health conditions and deaths. It is crucial to detect misinformation related to vaccination in social media. Here, we study anti-vaccination campaigns in Twitter and develop predictive models which could detect anti-vaccination tweets with high accuracy. This is challenging since the number of the anti-vaccination tweets are typically very less and results in highly imbalanced data. We develop a predictive model to take care of this data imbalance based on synthetic minority undersampling techniques which could overcome these limitations and detect anti-vaccination tweets with high accuracy and a very low False Positive rate. © 2019 IEEE.