Small datasets, missing values and the presence of multicollinearity often plague samples used in marketing research. Partial-least squares regression (PLS) is a statistical tool specifically designed to cope with such problems. The paper proceeds first with a concise review of the PLS method. It then highlights particular conditions when applying a standard OLS regression leads to unstable statistical results. The argument is supported with an empirical application on the market orientation/performance link. In conclusion, it is called for an increased reliance on the use of PLS in research.