Data on orienting and habituation to irrelevant sound can distinguish between task-specific and general accounts of auditory distraction: Distractors either disrupt specific cognitive processes (e.g., Jones, 1993; Salamé & Baddeley, 1982), or remove more general-purpose attentional resources from any attention-demanding task (e.g., Cowan, 1995). Tested here is the prediction that there is no further auditory distraction effect on immediate serial recall with increments in the number of distractors beyond the "changing-state point" of two discrete distractors. A Bayes factor analysis refutes this nil hypothesis: This prediction, a key element of the strong changing-state hypothesis, is shown to be less likely than two competing alternatives. Quantitative predictions for distraction as a function of the number of distracters are derived for an orienting-response (OR) and a stimulus-mismatch (SMM) hypothesis, representing general and task-specific accounts respectively. The data are shown to be more likely under the SMM hypothesis. Prospects for a parametric account of auditory distraction are considered.