Multivariate techniques summarize large sets of data and present their major trends in a graphically simplified manner. This paper discusses ways in which these techniques can be used to correlate biostratigraphical data. Stratigraphically constrained cluster analyses, which highlight distinct zones, may be performed on a number of sections. These can be plotted together to provide a concise summary that allows sequences to be compared. Ordinations, such as principal components and detrended correspondence analyses, can be performed on several sections simultaneously to extract a new set of axes that represent the combined trends of various taxa. The first few major axes can be plotted against depth as ‘biostratigraphical logs’; the rest can be ignored as insignificant variation or noise. Sections can then be correlated based on these logs, either manually or using procedures such as sequence slotting. These techniques have been tested using artificial data from Edwards (1984), consisting of taxa of known spatial and temporal ranges. Correlations between sections were found to match closely the true ages. Inconsistencies in the data, such as time-transgressive taxa, were isolated on one axis, allowing them to be removed or studied further.