Nuclear magnetic resonance (NMR) metabolomics profiling was evaluated as a new tool in sensory assessment of protein hydrolysates. Hydrolysates were produced on the basis of different raw materials (cod, salmon, and chicken), enzymes (Food Pro PNL and Bromelain), and hydrolysis time (10 and 50 min). The influence of raw material and hydrolysis parameters on sensory attributes was determined by traditional descriptive sensory analysis and 1H NMR spectroscopy. The raw material had a major influence on the attribute intensity and metabolite variation, followed by enzyme and hydrolysis time. However, the formation of bitter taste was not affected by the raw material. Partial least-squares regression (PLSR) on 1H NMR and sensory data provided good models (Q2 = 0.55−0.89) for 11 of the 17 evaluated attributes, including bitterness. Significant metabolite−attribute associations were identified. The study confirms the potential prediction of the sensory properties of protein hydrolysates from cod, salmon, and chicken based on 1H NMR metabolomics profiling.