The European Directive on drinking water quality has included mCP agar as the reference method for recovering Clostridium perfringens from drinking waters. In the present study, three media (mCP, TSCF and CP Chromo Select Agar) were evaluated for recovery of C. perfringens in different surface water samples. Out of 139 water samples, using a membrane filtration technique, 131 samples (94.2%) were found to be presumptively positive for C. perfringens in at least one of the culture media. Green colored colonies on CP Chromo Select Agar (CCP agar) were counted as presumptive C. perfringens isolates. Out of 483 green colonies on CCP agar, 96.3% (465 strains, indole negative) were identified as C. perfringens, and 15 strains (3.1%) were indole positive and were identified as Clostridium sordellii, Clostridium bifermentans or Clostridium tetani. Only 3 strains (0.6%) gave false positive results and were identified as Clostridium fallax, Clostridium botulinum, and Clostridium tertium. Variance analysis of the data obtained shows statistically no significant differences in the counts obtained between media employed in this work. The mCP method is very onerous for routine screening and bacterial colonies could not be used for further biochemical testing. The colonies on CCP and TSCF were easy to count and subculture for confirmation tests. TSCF detects sulfite-reducing clostridia, including species other than C. perfringens, and in some cases excessive blackening of the agar frustrated counting of the colonies. If the contamination was too high, TSCF did not consistently produce black colonies and as a consequence, the colonies were white and gave false negative results. On the other hand, the identification of typical and atypical colonies isolated from all media demonstrated that CCP agar was the most useful medium for C. perfringens recovery in water samples.