Feces from cattle fed 65 different diets were used to develop near-infrared spectroscopy (NIRS) calibrations to predict fecal composition [organic matter (OM), starch, nitrogen (N), neutral detergent fiber (NDF), acid detergent fiber (ADF), acid detergent lignin (ADL), and ether extract (EE)], and feces from cattle fed 23 different diets were used to predict digestibility [dry matter (DM), OM, starch, crude protein (CP), NDF, ADF, and gross energy (GE)]. The predictions of these parameters in feces collected from cattle fed wheat or barley grain (89% of diet DM) with barley silage (study 1) or barley grain diets containing varying levels of barley silage (study 2) were assessed. Validations for fecal composition displayed moderate to excellent linearity (R²_val ≥ 0.70) for OM, starch, N, NDF, and ADL and poor linearity for ADF and EE (R²_val = 0.25). When both digestibility and feedlot data sets were combined, cross-validation for fecal composition found accurate predictions (R²_CV ≥ 0.90, SECV ≤ 2.42) for all fecal constituents except EE. Accuracy of predicting digestibility was high for starch (R²_CV = 0.84, SECV = 1.06), good to moderate for DM, OM, CP, and GE (R²_CV ≥ 0.62, SECV ≤ 3.63) but poor for NDF and ADF (R²_CV ≤ 0.33, SECV ≥ 7.86). We conclude that fecal NIRS can predict both chemical composition and digestibility of most nutrients in feedlot cattle feces.