Natural compounds represent a significant source for the development of novel medicines. Finding the target proteins for a natural compound is the most important step towards understanding its molecular mechanism for therapeutic usage. In fact, the search for target proteins could be considered the first step of the drug discovery and development pipeline. While experimental determination of compound-protein interactions remains very challenging, effective in silico approaches have been developed and have demonstrated appealing advantages, including their low-cost and capability to scale up easily. The goal of this article is to provide an introduction to in silico search for drug targets of natural compounds. I first review currently available natural compounds databases and human gene/protein databases, and the rapidly emerging databases for known drug-target interactions. These resources provide the 'materials' for in silico approaches and define the gold standard of 'positives' for evaluating them. I then introduce three classes of computational methods for target identification of natural compounds, namely molecular docking, quantitative structure-activity relationship (QSAR) modeling, and data mining and integrative analysis. Use of these methods is explained using real examples, and the advantages and disadvantages of each method are compared. As these state-of-the-art methods continue to mature amid significant challenges, this field appears poised for a period of significant growth, with untold benefits to drug discovery and natural product development.