Results obtained with the standard repetition-change paradigm of orienting research cannot be attributed unambiguously to either stimulus change or to unexpectedness. By adding announcement conditions, in which participants were told about an impending stimulus change, these two factors were disentangled. In Experiment 1, reaction times (RTs) were longer and ratings of surprise were higher with unannounced than with announced stimulus change. In contrast, larger skin conductance response (SCR) magnitudes occurred following change, irrespective of its congruence with participants' expectations. Experiment 2 replicated the results for SCR magnitude and, furthermore, revealed the same pattern of results for the evoked cardiac response. Surprise ratings again reflected the unexpectedness of stimulus presentations. The dissociation between RT and autonomic measures provides difficulties for resource allocation accounts of the orienting response.