Honey mesquite (Prosopis glandulosa Torr.) is a common invasive shrub whose leaves contain secondary compounds that limit consumption by cattle. However, crude protein and fiber levels of mesquite leaves are similar to alfalfa hay. The ability to consume small quantities of mesquite leaves might make cattle more adaptable to southwestern rangelands. The objective of this study was to determine if consumption of mesquite leaves by cattle could be detected using fecal near infrared reflectance spectroscopy (NIRS). Six cannulated cows were fed either a cool season, warm season or cool and warm season mix grass hay ad libitum. Dry ground mesquite leaves were placed directly in the cannula to achieve 0, 1, 3 and 5% of the diet. Each level of mesquite was fed for 7 days and then increased to the next concentration. Fecal samples were collected on day 6 and 7, dried, ground, and analyzed using NIRS. Samples were separated into calibration (warm and cool season grass) and validation data sets (mix). Hay intake dropped dramatically for 3 of the 6 cows at the 5% mesquite level. No problems with intake or behavior were evident when mesquite comprised 3% or less of the diet. Fecal NIRS under-predicted mesquite in diet at 0 and 1% levels when two 0% mesquite samples from the validation data set were included in the calibration and over-predicted mesquite at all levels without validation samples in the calibration. Including the 0% mesquite fecal samples reduced the bias at higher mesquite levels (3 and 5%) by about 1 percentage unit compared to calibrations without fecal samples from the 0% diet. Simple coefficient of determinations did not differ (P > 0.50) between the two calibrations and were about 0.70. Fecal NIRS may have the ability to estimate the amount of mesquite leaves consumed by cattle grazing rangelands.