A 3-year experimental study was conducted on the factors influencing the use of paper-mill sludges on agricultural land. An array of 6 m × 6 m plots was used, allowing spreading to be carried out by full-size agricultural machinery. The factors investigated were crop type (grass and wheat), soil type (sandy loam and clay), sludge type (sludges from two different paper mills, designated A and D), method of sludge application (surface spreading of dewatered sludge, surface spreading of undewatered sludge and subsurface injection of undewatered sludge) and rate of sludge application (5 and 10 t DM/ha in the first season and 10 and 20 t DM/ha in the second and third seasons: in all cases applied each September). Soil condition was assessed in detail both at the beginning of the experiment before any sludge was spread and after the end of the third season. All plots received mineral nitrogen fertiliser in amounts according to normal farming practice, i.e. no ‘credit’ was given for any N, P or K content of the sludge. Several statistically significant differences in crop yield were found. The most pronounced difference was that Sludge A gave greater yields than did Sludge D. This effect was found for two seasons (1992 and 1993) for grass on clay soil as well as for one season (1992) for winter wheat on sandy loam soil (although in this last case yields were abnormally low). The cause of this effect was believed to be chemical differences between Sludges A and D, but they must be subtle ones because the only major chemical differences evident from the analyses were in the C N ratios, the potassium and the lead contents, and in all three cases there was evidence that these particular chemical differences were not the cause of differences in crop yield. The second most pronounced effect was found with the wheat on clay soil. In all three seasons, there was a significant effect on yield of method of sludge application. The level of significance differed from year to year: in 1991 and 1993 it was weak, but in 1992 it was very strong. In all three seasons, surface spreading of undewatered sludge gave the highest yields. The physical reason behind this effect may be differences in the distribution of sludge in the topsoil after sludge spread by the three different methods had been ploughed in prior to sowing the wheat. In most cases, topsoil condition, as assessed by the percentage of organic carbon, was significantly improved as a result of mill-sludge applications over 3 years. For example, grass plots on clay soil which received the higher rates of sludge application by injection had at the end of the experiment 2.99% organic carbon (Sludge D) and 3.15% organic carbon (Sludge A), while for the corresponding control plots, duplicate measurements gave 2.60 and 2.63%. In other, physical, tests to assess topsoil condition (e.g. measuring vol% air capacity, vol% total pore space and vol% total available water), it was found that, on the plots selected (all of which were wheat plots), spreading mill sludge over 3 years significantly improved the topsoil condition for the clay soil, but hardly did so on the sandy loam soil. The general conclusions from this work were that although a number of statistically significant differences in crop yield were found, none of these was large in absolute terms, and that significant improvements in topsoil condition were found in most cases, particularly when this was assessed by measuring the percent organic carbon. Therefore, when developing any strategy for persuading farmers to accept paper-mill sludges on their land, the case should be based mainly on the potential improvements in soil condition, perhaps arguing that increases in crop yield should follow from successive improvements in soil condition over a number of years. Such a benefit ought in due course to be particularly evident in very dry seasons.