Effective dog training depends on timely delivery of rewards. Critical to an understanding of "timely" for dogs is an understanding of the psychophysical performance of dogs when temporal durations are used as stimuli. Fifteen dogs were tested, food was used as reinforcers, and owners were asked not to feed their dogs before testing. The dogs were shown a white light for either a short or long duration, on the centre of the display. They were trained to nuzzle the lever above the screen whose colour is associated with the duration shown, for example touch the red screen lever when it was a 1 sec duration and the green screen lever when it was 4 sec. The location of the comparison stimuli was randomised across the left and right sides. Across the pairs of delays, difference in delays was one to four. If dogs made the correct response a piece of food is delivered. Once the dog responded with above 80% accuracy the testing phase began. During testing the two original signal durations were presented on 25% of trials on the remaining trials intermediate duration e.g. 6 sec are presented. Psychometric functions from stimulus-generalization sessions, when novel test durations were introduced, are presented.