The additives (urea, biuret and poultry litter) present in alfalfa, which contribute non-proteic nitrogen, were analysed using near infrared spectroscopy (NIRS) technology together with a remote reflectance fibre-optic probe. We used 75 samples of known alfalfa without additives and 75 samples with each of the additives, urea (0.01-10%), biuret (0.01-10%) and poultry litter (1-25%). Using the discriminant partial least squares (DPLS) algorithm, the presence or absence of the additives urea, biuret and poultry litter is classified and predicted with a high prediction rate of 96.9%, 100% and 100%, obtaining the equations of discrimination for each additive. The regression method employed for the quantification was modified partial least squares (MPLS). The equations were developed using the fibre-optic probe to determine the content of urea, biuret and poultry litter with multiple correlation coefficients (RSQ) and prediction corrected standard errors (SEP (C)) of 0.990, 0.28% for urea, 0.991, 0.29% for biuret and 0.925, 2.08% for poultry litter. The work permits the instantaneous and simultaneous prediction and determination of urea, biuret and poultry litter in alfalfas, applying the fibre-optic directly on the ground samples of alfalfa.