A field study was undertaken to compare dissolved organic carbon (DOC) concentrations in soil solutions obtained with three different sampling methods over a range of soil types. The sampling devices used were a tension-free collector, a tension Prenart collector and a tension Rhizon collector. Samples were collected fortnightly for a year at seven sites in northern England, each collection being replicated three times. The soil solution DOC ranged from 1.3 g m−3 in an acid ranker to 34.7 g m−3 in a peat. The DOC concentrations obtained with the three methods correlated reasonably well (r2 = 0.6–0.8) but with an indication of bias, as the best fit line differed from the 1:1 line. The tension-free collector gave generally higher DOC concentrations except at very low concentrations (in the acid ranker soil). The DOC concentrations measured with the tension-free collectors were significantly (P < 0.05) higher than those obtained with Prenart and Rhizon collectors at four and six sites, out of seven, respectively. Subsequent laboratory tests on tension-free collected samples showed no DOC loss on filtration through 0.1 and 0.22-μm membranes, whereas a significant loss of DOC occurred when tension-free collected samples were subsequently passed through Prenart and Rhizon collectors, indicating a probable sampling artefact with the tension devices. The difficulties of acquiring representative soil solution samples are discussed, together with the advantages and disadvantages of tension and tension-free methods.