Trees provide myriad ecosystem services of benefit to urban populations; however, urban development is pressuring existing urban tree coverage. Thus, a pertinent challenge for planners is identifying development scenarios that find synergies between urban growth and the preservation or enhancement of tree canopy coverage. This paper presents the training and validation of a model that predicts changes in neighbourhood-level urban tree canopy cover associated with different socio-economic and physical urban form variables. Neighbourhoods across Perth, the capital city of Western Australia, were used as a case study. A Random Forests model was trained using a suite of socio-economic and urban form variables and neighbourhood percentage tree canopy cover derived from very high resolution multispectral remote sensing images and digital surface models. This model was validated using independent test data with a mean absolute error of 1.78% and a root mean square error of 2.42%. An application of this model was demonstrated using the City of Nedlands, Perth, where a new planning scheme allowing denser urban development has been approved by the State Government. The magnitude and spatial variation in the change of neighbourhood tree canopy cover in the City of Nedlands in 2050 associated with three urban development scenarios was predicted using the model.