The economic, social and environmental benefits of urban green space are increasingly recognized by governments and urban planners. However, efficient methodologies of quantifying and monitoring the state of urban green spaces are needed to keep pace with urban development. The increased availability of high resolution satellite imagery can provide a means for the classification of urban green space and could potentially reduce the cost and time required to identify green space compared to the present approach of visually interpreting aerial photographs. In the past, traditional pixel-based approaches to image classification have struggled to produce satisfactory results for the classification of urban green spaces. Object-oriented image analysis may provide a means to classify meaningful image objects by making greater use of the spectral, spatial, textural and contextual properties of high resolution imagery. In this study, SPOT-5 images were used to develop a fuzzy membership rule-set for the classification of urban green space in Kuala Lumpur, Malaysia. Four broad categories of land cover were segmented and classified at two different scales. Although there was some confusion between vegetation classes at finer segmentation, overall user accuracy for extracting Grassland, Trees, Sports Fields and Square Lots was satisfactory with accuracies between 70%-100% at most sites. Results suggest that object-oriented image analysis can be an effective tool for classifying broad distinctions between green and grey land cover types from high resolution satellite imagery, however improvement in rule-set development will be needed before the more detailed green space categories can be reliably separated.