Accurate inventories of grassland are important for studies of carbon dynamics, biodiversity and agricultural management. In Ireland, grassland is the dominant land cover, occupying approximately 64% of the country`s land area and representing over 90% of all agricultural land (~4,000,000ha). Due to persistent cloud cover and considering upcoming missions (Sentinel-1 & ALOS-2), the use of multi-temporal synthetic aperture radar (SAR) data provides an attractive solution for providing up-to-date inventories of grasslands and provides opportunities for identifying and monitoring changes. In this study, and as part of the wider Irish Land Mapping Observatory (ILMO) project, the performance of non-parametric classifiers (Random Forests and Support Vector Machines) in discriminating elements of the Irish landscape is investigated using multi-temporal SAR data. The study sites cover two counties located in the midlands and North West of Ireland and the EO-database consists of ENVISAT ASAR, ERS-2 and ALOS PALSAR data acquired throughout 2008. For comparative purposes, the results are also compared to those from a traditional statistical classifier. In a final step, the results are used to populate the Ordnance Survey of Ireland`s (OSi) Prime2 spatial boundary database to produce a seamless thematic output that accurately delineates real-world parcels. If carried out on a national scale, the classifications could contribute to future assessments of Ireland`s greenhouse gas (GHG) inventory for the (extended) Kyoto protocol (2013-2020), EU reporting and other national assessment requirements.