Drug discovery is one of the research areas that is most critical for human health and well-being because of the tremendous impact of safe and efficient drugs on humans. Despite active research, the number of newly discovered drugs has been low in the last few decades because biological systems are complex and experiments are expensive. To increase the efficiency of the drug discovery process, in-silico screening has become a standard approach, using computational methods as efficient surrogates for wet lab experiments. Among these methods, AI methods have been found to be one of the most promising candidates, but they often require large amounts of data while, for many drug discovery projects, available data is scarce. Therefore, few-shot learning emerged as a subfield of AI, which refers to scenarios where the amount of data is limited and to models specifically designed for such low-data scenarios. This chapter introduces the concept of few-shot learning in the context of drug discovery and provides an overview of recent advancements aimed at overcoming the challenges posed by data scarcity.