SUMMARYFour soils of different texture (6% to 46% clay and 6% to 32% silt) were incubated with14C‐labelled barley straw for periods ranging from 16 to 18 years. Incubations were either under laboratory (three soils, 20°C) or field conditions (one soil).Soil samples were separated into clay‐size (<2 μm), silt‐size (2–20 μm) and sand‐size (20–2000 μm) fractions using ultrasonic dispersion and gravity sedimentation. Whole soils and size fractions were hydrolysed with 12 M sulphuric acid, the released neutral sugars isolated and their amounts and specific activities determined.The proportion of C present as carbohydrate decreased in the order clay, silt and sand, but the sugar compositions of the various fractions were similar. No relationship could be discerned between the soil particle‐size distributions and the persistence of sugars.The specific activity of xylose remained higher than that of any major sugar, indicating the persistence of straw polysaccharides. This was true for all soils and size fractions. Galactose and mannose retained a greater proportion of their initial activity than did arabinose, xylose or glucose. This was ascribed to synthesis of microbial polysaccharides.Total activity present in whole soils decreased in the order glucose, xylose and arabinose, galactose and mannose. The total activity of individual sugars in size fractions decreased in the order clay, silt, sand. The hexose to pentose ratio (galactose+mannose to arabinose + xylose) tended to be higher for silt than for clay, but the trend was not consist