We hypothesize, based on management control processes in large firms (Covaleski et al., 1998), that large-firm practitioners will be less likely than small-firm auditors to report beliefs that non-audit services (NAS) impair auditor independence. Based on Goldman and Barlev's (1974) analysis of auditor-firm conflict of interests and procedural independence safeguards, we also predict that auditors will report less concern over impairment than non-auditors. We investigate also belief perseverance - whether practitioners tend to maintain prior reported beliefs after reading research on the relationship between NAS and auditor independence. Practitioners reported their beliefs before and after reading a summary of Frankel et al. (2002) indicating that providing NAS may impair independence; Ashbaugh et al. (2003) suggesting that independence is not impaired; or a summary of both findings. Twenty-six percent of small-firm practitioners and 23% of non-auditors reported prior beliefs that NAS impair independence, much larger proportions than for large-firm practitioners (5%) and auditors (6%). After reading the summary 20% of small-firm practitioners and 19% of non-auditors changed their answers, larger proportions than for large-firm practitioners (9%) and auditors (8%). Results generally support our hypotheses and suggest that as evidence regarding the effectiveness of key provisions of the Sarbanes-Oxley Act of 2002 emerges, accounting practitioners will maintain their beliefs, but react differently depending upon their specializations and firm affiliations. This will affect public policy to the extent that accountants and auditors, rather than academics, report research findings to regulators.