The effectiveness of cycloserine-cefoxitin-fructose agar (CCFA) and taurocholate-CCFA (TCCFA) in isolating Clostridium difficile from swabs of the rectum or stools from 184 hospitalized patients who were monitored weekly when they had diarrhea was compared. The number of surveillance time points ranged from two to eight per patient over a period of 4 to 34 days per patient, totalling 621 comparisons of the media. C. difficile was isolated more frequently by TCCFA than CCFA at seven of eight surveillance points, a significant trend (O'Brien test, p = 0.002). This difference reached statistical significance at the second surveillance time point when the prevalence of C. difficile was sufficiently high. At the second surveillance point, C. difficile was isolated only by TCCFA in 7 of 184 comparisons of the media, only by CCFA in none of the comparisons, and by both median in 19 comparisons (p = 0.016). C. difficile was first isolated at an earlier surveillance time point on TCCFA in 11 of 36 patients and on CCFA first only once (p = 0.005). Use of TCCFA media increased the rapidity and sensitivity of culture for C. difficile when doing patient surveillance but did not increase sensitivity when diagnosing patients with diarrhea.