Registration of complications of treatment is an important instrument for measuring the quality of health care. Reliable registration depends on definitions, the case-finding method that is used, and the registration method itself. We conducted a comparative study of two different methods of registration for the surveillance of surgical site infections (SSIs) in a single hospital. The study included all patients in both the surgical database and the microbiology and infection-prevention database of the hospital who underwent surgery on the abdominal aorta or peripheral vascular procedures from March 1, 2009 to March 1, 2010. The surgical database included positive scores for SSI in cases of positive wound swabs, the need for incision drainage, or the need for antibiotic treatment. The microbiology and infection-prevention database used criteria from the U.S. Centers for Disease Control and Prevention (CDC), and based positive scores on redness, heat, swelling, or pain in the area of a surgical incision within 30 d after a procedure, and on a positive swab, drainage from an incision, or the presence of pus following a diagnostic puncture. The surgical complication database included 218 patients, of whom 20 (9.2%) had a SSI. The microbiology and infection-prevention database included 236 patients, of whom 33 (14%) had a SSI. The databases were merged and all infections were ascertained by an expert team. The surgical database had a sensitivity of 57% for SSIs, whereas the microbiology and infection-prevention database had a sensitivity of 93% (p<0.05). Physicians provided less reliable scores for SSI than did trained infection-control practitioners. This raises questions about the comparability of rates of SSI in different institutions as a means for judging the quality of hospital care.