Anaerobically grown cells of Escherichia coli were immobilised within a range of entrapment matrices and packed into a column under standard conditions, and the ability of the immobilised cells to reduce nitrite (0.5 mM) was measured at a range of flow rates using sodium formate (20 mM) as the electron donor for nitrite reduction. A flow-rate/activity plot was constructed for each flow-through reactor and RA(1/2) values (residence lime corresponding to 50% nitrite removal) calculated for each reactor type. Cells immobilised in flat and hollow-fibre membranes were the most effective (RA(1/2) = 0.35 h and 0.47 h respectively), with cells entrapped by dialysis membrane (1.53 h), alginate beads (1.93 h). Hypol foam (2.31 h) and polyacrylamide gel (50% nitrite not removed at maximum residence time tested : 4.9 h) performing progressively less effectively. Cells grown as a biofilm on a range of support materials were also tested in comparable packed-bed reactors. Cell loss from these supports was extensive and contributed to poor performance of the reactors despite high initial biomass loadings (RA(1/2) values using raschig rings, coke and activated-carbon supports : 1.6 h, 2.3 h and 1.0 h respectively). Biofilms grown on Pharmacia microcarrier supports and used in packed and also fluidised beds were more stable and the performance of these reactors was superior to that of biofilm reactors using other supports, and comparable to that of the membrane reactors (RA(1/2) values for Cytoline 2, Cytopore 2 and Cytodex 3 : 0.76 h, 0.56 h, 0.68 h respectively).