Testate amoebae are useful environmental indicators in ecological and palaeoecological studies from peatlands. Previous quantitative studies have focused on the peatlands of Northern and Central Europe, North America, and New Zealand and have considered a relatively restricted variety of peatland types, mostly ombrotrophic or Sphagnum-dominated while more minerotrophic fens have been less studied. Here we present the first quantitative ecological study of testate amoebae from four small mesotrophic fens (pH 5.5-8.1) in the Elatia Forest, northern Macedonia province, Greece. Relationships with the environmental data were investigated using redundancy analysis and mantel tests. Transfer function models were derived using a variety of techniques. Results demonstrate that as for Sphagnumdominated mires hydrology is the most important control on amoebae community composition. Transfer function models should enable water tables to be predicted within 2.5 cm, when data selection is used this is reduced to less than 2 cm. pH is also an important environmental control on testate amoebae communities, a transfer function model enables pH prediction within 0.4 pH units. The hydrological transfer function is the best performing such model yet produced in terms of prediction error. This study provides new data on the ecology of testate amoebae in fens, and the transfer function models should allow quantitative palaeohydrological reconstruction.Key words: protists; transfer functions; palaeoclimate; peatlands; mire surface wetness.Abbreviations: Boot, results assessed by bootstrap cross validation e.g. RMSEP boot denotes RMSEP assessed by bootstrapping; DWT, depth to water table; Jack, results assessed by jack-knife (also termed 'leave-one-out') cross validation e.g. RMSEP jack denotes RMSEP assessed by jack-knifing; LOI, loss on ignition; MaxBias; maximum bias; ML, maximum likelihood; PLS, partial least squares; RMSEP, root mean square error of prediction; WA, weighted average; WA-PLS, weighted average partial least squares.