Understanding neuronal networks is very important for understanding the operational principles of the brain, such as how the brain encodes information and performs computations. Using various measurements of neuron activity, how do we infer interactions among them? In this talk, we will be focusing on different ways of connectivity inference for neural recordings. We will first review model-free and model-based methods including supervised learning approach, generalized linear model, dynamic Bayesian network, etc. For model-based methods, we will also talk about estimation of model parameters. Finally, we will discuss about the challenge for connectivity inference and future directions.