The protozoan parasite Cryptosporidium parvum is known to occur widely in both source and drinking water and has caused waterborne outbreaks of gastroenteritis. To improve monitoring, the U.S. Environmental Protection Agency developed method 1622 for isolation and detection of Cryptosporidium oocysts in water. Method 1622 is performance based and involves filtration, concentration, immunomagnetic separation, fluorescent-antibody staining and 4???,6-diamidino-2-phenylindole (DAPI) counterstaining, and microscopic evaluation. The capsule filter system currently recommended for method 1622 was compared to a hollow-fiber ultrafilter system for primary concentration of C. parvum oocysts in seeded reagent water and untreated surface waters. Samples were otherwise processed according to method 1622. Rates of C. parvum oocyst recovery from seeded 10-liter volumes of reagent water in precision and recovery experiments with filter pairs were 42% (standard deviation [SD], 24%) and 46% (SD, 18%) for hollow-fiber ultrafilters and capsule filters, respectively. Mean oocyst recovery rates in experiments testing both filters on seeded surface water samples were 42% (SD, 27%) and 15% (SD, 12%) for hollow-fiber ultrafilters and capsule filters, respectively. Although C. parvum oocysts were recovered from surface waters by using the approved filter of method 1622, the recovery rates were significantly lower and more variable than those from reagent grade water. In contrast, the disposable hollow-fiber ultrafilter system was compatible with subsequent method 1622 processing steps, and it recovered C. parvum oocysts from seeded surface waters with significantly greater efficiency and reliability than the filter suggested for use in the version of method 1622 tested.