Multivariate analysis is appropriate whenever more than one variable is measured on each sample individual, and overall conclusions about the whole system are sought. Many different multivariate techniques now exist for addressing a variety of objectives. This brief review outlines, in broad terms, some of the more common objectives and sketches out some of the ways in which they are tackled. The techniques are grouped under four headings: visualization and description, extrapolation and inference, discrimination and classification, modeling and explanation. Within each heading, the chief objectives are first identified, and the rationales behind some of the specific techniques are sketched out. The review ends with an outline of current trends and recent developments.