The cultural specificity of action prediction was assessed in 8-month-old Chinese and Swedish infants. Infants were presented with an actor eating with a spoon or chopsticks. Predictive goal-directed gaze shifts were examined using eye tracking. The results demonstrate that Chinese infants only predict the goal of eating actions performed with chopsticks, whereas Swedish infants exclusively predict the goal of eating actions performed with a spoon. Infants in neither culture predicted the goal of object manipulation actions (e.g., picking up food) performed with a spoon or chopsticks. The results support the view that multiple processes (both visual/cultural learning and motor-based direct matching processes) facilitate goal prediction during observation of other peoples' actions early in infancy.