Some errors crept into this Primary Care paper by Anders Beich and colleagues (BMJ 2003;327: 536-42). In the results section we failed to spot a small inconsistency between the text and table 4. In the first paragraph (abridged and printed version; second paragraph in the full, web version) of the section “Intervention effect and assessment efforts,” we said “NNTs [numbers needed to treat] of single studies ranged from 5 to 61,” but the correct lower level of this range should be 6, as table 4 shows. Additionally, we inexplicably published a few wrong values in table 5 (full version). The screening effect in the study by Fleming (ref 42) is 3.7 [not 0.7], and the maximum number of drinks for women in the studies by Ockene and Fleming (ref 41) is 4 [not 3] and for men in the study by Anderson is 11 [not 5].