Pigeons were trained in a choice matching-to-duration task in which the samples were 2 and 8 s in duration in Exp. 1, and 1 and 30 s in duration in Exp. 2.  Following training with variable delays ranging from 1 to 3 s, testing at delays of 2, 15, and 30 s revealed a choose-short effect in both experiments.  Both experiments involved a second training phase in which the range of variable delays was extended to 1 to 30 s.  A subsequent test, again involving delays of 2, 15, and 30 s, revealed systematic error tendencies (i.e., choose-short and choose-long effects) only in pigeons trained with 1 and 30 s samples.  The results are consistent with the view that systematic error tendencies in the matching-to-duration task are mediated by processes of analogical coding and subjective shortening.