The excessive expansion of urbanized areas has resulted in haphazard land utilization, immoderate consumption of superior agricultural land and water resources, significant fragmentation of agricultural landscape, and gradual deterioration of the agro-ecological environment. Combined, these factors cause poor land use efficiency. Under these circumstances, comprehensively assessing land use efficiency for urban agriculture is a key issue in land use research. Currently, evaluation methods for agricultural land use efficiency narrowly concentrate on aspects of economic input and output. However, urban agro-ecosystems can provide diverse economic, social, and ecological services and functions. In particular, the social and ecological services and functions originating from agricultural land, which have a higher value than economic services, play a significant role in ensuring regional social, ecological, and environmental security. However, recent research has rarely taken these benefits into consideration. Therefore, land use value has been greatly underestimated, which has resulted in mishandled and poor land use policies. In this study, we apply Landsat imagery and social and economic statistical data for the Xi’an metropolitan zone (XMZ) to investigate agricultural multi-functionality. We develop an evaluation framework for urban agricultural land use efficiency and identify agro-ecosystem services and functions as important outputs from agricultural land. The land use efficiency of urban agriculture is then evaluated using ecosystem services models, providing a mechanism for assessing spatial-temporal changes in land use efficiency in the XMZ from 1999 to 2015. Four important conclusions are reached from this analysis. First, the rapid urbanization and agricultural transformation from traditional cereal cultivation to modern urban agriculture has resulted in steadily increasing costs, outputs, and land use efficiency of urban agriculture. The total output value increased 41% and land use efficiency per hectare increased by 33.13% on average. Second, the spatial patterns of comprehensive output and land use efficiency were dominated by economic outputs from agricultural land. Areas near cities, which are dominated by orchard and arable land, provide more economic functions. These areas support and regulate services due to the transformation from extensive cereal production to intensive modern urban agriculture; therefore, they have higher output value and land use efficiency. In contrast, areas distant from cities, towns, and high traffic roads, namely, remote rural areas, provide more support and regulating services, but have relatively lower economic function due to inaccessibility to urban markets and slow agricultural transformation. Therefore, these areas have lower output value and land use efficiency. The spatial change in agricultural output and land use efficiency in urban areas is strongly dependent on the degree of urbanization and agricultural transformation. Third, the total output value and land use efficiency of urban agriculture measured with our approach are much higher than evaluations using traditional methods. However, the spatial patterns measured using the two approaches are in agreement. The evaluation framework integrates ecological services and economic and social functions into a comprehensive output from agricultural land. This approach is more methodical and accurate for evaluating the comprehensive efficiency of land use based on quantities and spatial scale because they are at the pixel scale. Finally, the evaluation results have important implications for enhancing current agricultural subsidies and even implementing ecological payment policies in China. Most importantly, they can be directly applied to agricultural transformation regulations, decision-making, and guidance for rational land utilization.