Social work researchers should become familiar with an econometric approach, known as instrumental variable estimation, to estimate causal effects when the variable of interest is not randomly assigned and is, therefore, nonignorable or endogenous. In this article, we introduce instrumental variable estimation and describe how this method is used in model estimation. We focus on the critical assumptions needed to support causal inferences using instrumental variable estimation, provide clear examples of instruments, and describe how these assumptions promote the validity of inference with instruments. We compare the instrumental variable approach to causal estimation with the path model, which is a familiar approach in most social work research, and provide a balanced appraisal of the advantages and disadvantages of each approach.