Several studies suggest that dogs, as well as primates, utilize a mental representation of the signaler after hearing its vocalization and can match this representation with other features provided by the visual modality. Recently it was found that a dogs' growl is context specific and contains information about the caller's body size. Whether dogs can use the encoded information is as yet unclear. In this experiment, we tested whether dogs can assess the size of another dog if they hear an agonistic growl paired with simultaneous video projection of two dog pictures. One of them matched the size of the growling dog, while the other one was either 30% larger or smaller. In control groups, noise, cat pictures or projections of geometric shapes (triangles) were used. The results showed that dogs look sooner and longer at the dog picture matching the size of the caller. No such preference was found with any of the control stimuli, suggesting that dogs have a mental representation of the caller when hearing its vocalization.