Quaternary temperate stages have long been described based on changing pollen abundances of various tree taxa in lacustrine sediments. Later, attempts have been made to assign such biostratigraphic units to distinct marine isotope stages (MIS). Existing continuous chronosequences from Southern Europe provide good chronologies and thus enable a biostratigraphic definition of at least younger MIS. In Northern Europe, however, the fragmentary character of the records and the weaknesses of absolute dating prevent good age estimates. Therefore, age-determination of the majority of fragmentary records depends on site-to-site correlations. This comparison has often been performed on a visual basis, lacking clearly defined protocols and statements of underlying assumptions. Here I test the correlation of well and poorly known pollen records of the middle- and late-Pleistocene temperate stages from Northern-Central Europe and evaluate the usefulness of several numerical techniques. TWINSPAN analysis identifies groups of temperate stages based on presence/absence of their indicative taxa and may be useful for distinguishing between older and younger interglacials. Site-to-site sequence slotting allows the determination of the most similar pairs of records, based on sample dissimilarity following their stratigraphic constraints. Sequence slotting performs well when correlating the Holsteinian interglacial and Cromerian stage II, and also provides tentative correlation of some problematic records. Ordination compares main trends in pollen stratigraphies of all pollen sequences. It finds very similar patterns between Eemian records and Cromerian stage II. Although different methods show sometimes inconsistent results, they can certainly contribute to the discussion of the age of poorly known interglacials. The implications for future directions suggest focusing on better sampling resolution, multi-proxy approaches to climatic reconstruction and obtaining better independent dating.