Background: It has been suggested that dogs display a secure base effect similar to that found in human children (i.e., using the owner as a secure base for interacting with the environment). In children, this effect influences their daily lives and importantly also their performance in cognitive testing. Here, we investigate the importance of the secure base effect for dogs in a problem-solving task. Methodology/Principal Findings: Using a manipulative task, we tested dogs in three conditions, in which we varied the owner's presence and behavior (Experiment 1: “Absent owner”, “Silent owner”, “Encouraging owner”) and in one additional condition, in which the owner was replaced by an unfamiliar human (Experiment 2: “Replaced owner”). We found that the dogs' duration of manipulating the apparatus was longer when their owner was present than absent, irrespective of the owner's behavior. The presence of an unfamiliar human however did not increase their manipulation. Furthermore, the reduced manipulation during the absence of the owner was not correlated with the dog's degree of separation distress scored in a preceding attachment experiment. Conclusions/Significance: Our study is the first to provide evidence for an owner-specific secure base effect in dogs that extends from attachment tests to other areas of dogs' lives and also manifests itself in cognitive testing – thereby confirming the remarkable similarity between the secure base effect in dogs and in human children. These results also have important implications for behavioral testing in dogs, because the presence or absence of the owner during a test situation might substantially influence dogs' motivation and therefore the outcome of the test.