This paper proposes an evaluation method of human stress and drowsiness using photoplethysmography (PPG) measured from ears through the use of earbud-type photoplethysmographic sensor. For improving robustness with respect to body motion, velocity pulse waves have been calculated to detect motion artifact. In addition, to increase the temporal resolution and cope with misalignments in sensor position, 4 simple features have been derived from the velocity pulse waves. These 4 features are selectively related to stress and drowsiness scores using multiple regression analysis and a stepwise method. The stress and drowsiness evaluation models were implemented and tested with the validation dataset collected in our experiments. In the simulation results, human stress and drowsiness were evaluated successfully from PPG measured from ears. As the results show, the proposed method enabled human stress and drowsiness evaluation using PPG measured through the earbud-type photoplethysmographic sensor.