A multivariate statistical method (factor analysis of mixed data and hierarchical classification) was used to classify the environmental settings where sorted patterned ground develops in a wet oceanic periglacial area (Skagafjörður, Northern Iceland). A total of 750 periglacial features, distributed over 75 sites, were studied. Nine explanatory variables were assessed by fieldwork and using a digital elevation model, the variables were subdivided into three groups (latitude, topography and soil characteristics) and then integrated into a geographical information system. Furthermore, a correlation between the environmental variables and an intrinsic variable (patterned ground mesh diameter) was determined by a bivariate test. The results show that sorted patterned ground are spread over three homogenous areas, mostly differentiated by altitude, insolation, grain size characteristics and type of drift. In addition, feature diameters differ significantly from one group to another. Finally, it appears that patterned ground diameters are positively correlated with (i) the proportion of clay to medium silt content ( r = 0.35), (ii) altitude ( r = 0.51), and especially with (iii) clast length ( r = 0.97). This strong relationship with clast length is observed in each homogenous patterned ground area at both site and feature scales.