Lack of research on farm-level and temporal stability has slowed the development of precision agriculture. The objectives of this study were to delineate management zones based on apparent electrical conductivity, elevation, Landsat imagery, and yield maps on a farm level and evaluate the stability of the management zones between seasons. This study was conducted in eight commercially managed cotton fields in the Southern High Plains. Apparent electrical conductivity data was collected using a Veris EC mapping system in 2001; elevation data was obtained using a real-time kinematic (RTK) system in 2004; and yield monitor data and two Landsat images for bare soil brightness were obtained from the year 2003 and 2004. These data attributes were aggregated based on the 30 by 30 m Landsat imagery grid. Three management zones for each field and each year based on these attributes were delineated using the SAS FASTCLUS procedure and Management Zone Analyst (MZA). Results indicated that management zones separated by Veris EC, lint yields, bare soil imagery and terrain attributes had consistent and logical groupings for fields with great variability in soil properties, e.g. high yielding zones/clusters had low elevation, slope, and bare soil brightness, and high EC. Management zones were relatively stable among the seasons in these fields. For other fields, however, the management zones by these attributes were not consistent between different seasons. The SAS FASTCLUS procedure had more logical and consistent management zone delineation than MZA. Farm level research helps determine the suitability of different fields for precision farming.