The large costs associated with modern drug discovery mean that governments and regulatory bodies need to provide economic incentives to promote the development of orphan drugs (i.e., medicinal products that are designed to treat rare disease that affect only small numbers of patients). Under European Union (EU) legislation, a medicine can only be authorised for treating a specific rare disease if it is not similar to other orphan drugs already authorised for that particular disease. Here, we discuss the use of 2D fingerprints to calculate the Tanimoto similarity between potential and existing orphan drugs for the same disease, and present logistic regression models correlating these computed similarities with the judgements of human experts.