Based on the analysis of urban construction land expansion and its driving mechanism, an econometric model to evaluate the level of urban land intensive use was established. The regional differences and the temporal-spatial distribution of the levels of urban land intensive use in Guangdong Province were measured quantitatively. Then the results were compared with that of the traditional index weighting method. The results show that the econometric model proposed in the paper can measure the level of urban land intensive use very well. Viewed from the elastic coefficient of socio-economic variables, the expansion of urban land is mainly affected by the increase of urban population and residents’ income, while the effect of economic growth and fixed asset investment is not obvious. But the rising price of land and the strict land management can effectively inhibit the sprawl of urban land. The level of urban land intensive use was improved in Guangdong Province from 2003 to 2012. But there existed serious waste of land use in some cities. The gaps of land intensive use levels among cities are obvious in Guangdong Province. As viewed from the overall evolution, the levels of intensive use of seven cities in Pearl River Delta and seven cities in west and east Guangdong have declined, while the levels of land use intensive of seven cities on the mountain area have increased. The traditional index weighting method may be more suitable for the evaluation of industrial land, development zone and other productive land. If it was used to make a comprehensive evaluation on all urban land, it may exaggerate the effect of output and investment on land intensive use. Because of the law of diminishing land returns, high input and yield may not be the most economical and effective land use pattern. The results based on Data Envelopment Analysis show that the land use pattern with high investment in exchange for high yield is not necessarily intensive utilization.