A convolutionally encoded data is usually decoded by applying the Viterbi algorithm to the associated code trellis. On the other hand, syndrome decoding based on error trellises was proposed by Schalkwijk et al. This paper mainly focuses on error trellises and syndrome decoding based on them. Error trellises have a kind of nonuniformity, which enables decoding with remarkably low average complexity. It is shown that scarce-state-transition (SST) Viterbi decoding based on a code trellis is equivalent to syndrome decoding based on the corresponding error trellis. Code/error trellises for tail-biting convolutional codes are also discussed in this paper.