Rupture of carotid plaques triggers stroke. Current diagnosis of stroke is based on lumen stenosis, resulting in sever overtreatment. Photoacoustic (PA) imaging can provide comprehensive and patient-specific assessment of plaque vulnerability, and prevent overtreatment. However, no in vivo PA imaging of carotid plaque is available due to low SNR. Here, we present a fast PA/US imaging system and motion corrected averaging algorithm to increase PA SNR. The imaging system and algorithm are verified ex vivo, and in vivo on patients during carotid endarterectomy (intra-operatively). The results may accelerate the clinical translation of PA imaging of carotid plaques.