In remote sensing imagery, various normalized difference indices are widely used for land cover mapping. Each index has its targeting cover type with a specialized data source. However, these indices are generally only studied in multispectral data. Hyperspectral images have become increasingly attractive due to their richness of spectrum information. A new index, i.e., Normalized Difference Built-up Index for Hyperspectral data (NDBIh), oriented to built-up land enhancement in hyperspectral remote sensing data is proposed. Spectral response curves of different cover types and possible calculation equations for NDBIh are obtained first. The equation having the best ability to differentiate built-up land from other areas is referred to as NDBIh. To evaluate the ability of our NDBIh, two other built-up indices, the conventional Normalized Difference Built-Up Index (NDBI) and the Index-based Built-Up Index (IBI), are compared with NDBIh both qualitatively and quantitatively. Experiments on airborne visible infrared imaging spectrometer data indicate that the NDBIh outperforms NDBI and IBI in identifying built-up land.