Cheddar cheeses were made from raw milk, pasteurised milk (72 degrees C, 15 s) or milk produced from skim milk which had been microfiltered using an Alfa-Laval MFS-1 MF unit and mixed with pasteurised cream (72 degrees C, 30 s). Microfiltration (MF) reduced the total bacterial count (TBC) by > 99% and MF cheesemilk had a lower TBC than pasteurised milk; counts of nonstarter lactic acid bacteria (NSLAB) were < 1/ml. Cheeses were ripened at 10 degrees C. Commercial graders and a trained taste panel found the pasteurised and MF cheeses to be of high and equivalent sensory quality but the raw milk cheese rapidly developed a strong and atypical flavour and was downgraded. NSLAB counts for the MF, pasteurised and raw milk cheeses were 43, 3.9 x 10(2) and 1.47 x 10(5) cfu/g after pressing and 9.3 x 10(6), 1.12 x 10(7) and 1.19 x 10(8) cfu/g after 11 weeks. The cheeses and their water-soluble extracts were indistinguishable by urea-polyacrylamide gel electrophoresis for up to 3 months, but slight differences were apparent at 6 months. Peptide profiles were similar for pasteurised and MF cheeses but different for the raw milk cheese, Water-soluble nitrogen levels in the cheeses were similar although amino acid nitrogen and free glutamic acid were consistently highest in the raw milk cheese. The raw milk cheese exhibited the highest concentration of free fatty acids throughout ripening. The non-starter microflora, consisting predominantly of Lactobacillus spp., differed in the raw and pasteurised cheeses. The results show that MF efficiently reduced the numbers of indigenous microorganisms in milk and cheese. The results also suggest that the indigenous microflora of milk markedly affects the quality of Cheddar cheese made from raw milk.