Dog training consists of the dog's response and correct action to verbal commands given by the trainer. This study investigated the effects of non-verbal signals and trainer position on the response of dogs during training to two verbal commands. Seven dogs (aged 17 to 96 months; five female, two male; three purebreeds) were trained individually by the same female trainer in six trials × three training series to two commands ("sit" and "come"), controlling for trainer posture and position. The trainer gradually distanced herself from the dog's standing position. Initial distance between trainer and dog was 70cm (trial 1); it was finally increased to 420cm (trial 6). The trainer changed her position in a straight line facing the dog. Each training series was organised into six trials. The first training series: hand signals and body language were used when the experimental trainer presented the command ("sit" or "come"). The second training series: hand signals and body language were restrained when the experimental trainer presented each command. The third training series was similar to that in the first training series. The percentage of correct responses to each command in a given session of a given training trial was recorded, together with the number of sessions required to reach the success criterion (85% correct response) before progression to the next trial. Training at a given trial was conducted in sessions of 40 exercises; the two commands were presented randomly 20 times each. Data collected in response to both commands were analysed on a percentage-correct basis, i.e. the percentage correct in a given session of a given training trial. There were statistically significant differences among the three training series in the number of training sessions required to learn the "sit" command (ANOVA: F[2,125]=11.02, P<0.001). In Training A, there was a significant increase in the number of errors between trial 1, in which the experimenter stood facing the dog at a distance of 70cm, and trial 2, in which the experimenter stood facing the dog at a distance of 140cm (w=27, P=0.02) for the "sit" command. The influences of both distance and non-verbal signals on dog performance varied with the type of command. The results suggested that the first training trial was more important for the "sit" than for the "come" command. This might have been related to the difference in distance between the trainer and the dog after the dog had responded to each of these commands.