This paper examines the potential of remote sensing–derived metrics of vegetation phenology and a Multi-Layer Perceptron neural network to model the most likely locations of large, agglomerated archaeological sites. Focusing on two different environments in central New Mexico, the Galisteo Basin and the Sandia-Manzano Mountain range, this pilot study distinguishes between archaeological sites and their surroundings based on differential growth in vegetation. Using data derived from Landsat Thematic Mapper, a time series of Normalized Difference Vegetation Indices were created to characterize vegetation phenology in the study areas. Distinguishing between archaeological sites and their surroundings, the neural network was trained on a series of known sites to develop an output activation layer indicating the possible locations of other, previously unknown sites. This output activation layer, treated as a site suitability model, was validated using the receiver operating characteristic area under the curve using known sites excluded from the training procedure. Results show promise in large, open areas such as basin environments. While differences in vegetation type have relatively little effect, differences in elevation, or more directly the changes in phenology that go along with them, negatively impact the ability to infer the presence of archaeological sites using this approach.