This study was carried out to estimate the prevalence and potential for human infectivity of Giardia cysts in Canadian drinking water supplies. The presence of Cryptosporidium oocysts was also noted, but isolates were not collected for further study. A total of 1,760 raw water samples, treated water samples, and raw sewage samples were collected from 72 municipalities across Canada for analysis, 58 of which treat their water by chlorination alone. Giardia cysts were found in 73% of raw sewage samples, 21% of raw water samples, and 18.2% of treated water samples. There was a trend to higher concentration and more frequent incidence of Giardia cysts in the spring and fall, but positive samples were found in all seasons. Cryptosporidium oocysts were found in 6.1% of raw sewage samples, 4.5% of raw water samples, and 3.5% of treated water samples. Giardia cyst viability was assessed by infecting Mongolian gerbils (Meriones unguiculatus) and by use of a modified propidium iodide dye exclusion test, and the results were not always in agreement. No Cryptosporidium isolates were recovered from gerbils, but 8 of 276 (3%) water samples and 19 of 113 (17%) sewage samples resulted in positive Giardia infections. Most of the water samples contained a low number of cysts, and 12 Giardia isolates were successfully recovered from gerbils and cultured. Biotyping of these isolates by isoenzyme analysis and karyotyping by pulsed-field gel electrophoresis separated the isolates into the same three discrete groups. Karyotyping revealed four or five chromosomal bands ranging in size from 0.9 to 2 Mb, and four of the isolates had the same banding pattern as that of the WB strain. Analysis of the nucleotide sequences of the 16S DNA coding for rRNA divided the isolates into two distinct groups corresponding to the Polish and Belgian designations found by other investigators. The occurrence of these biotypes and karyotypes appeared to be random and was not related to geographic or other factors (e.g., different types were found in both drinking water and sewage from the same community). Biotyping and karyotyping showed that isolates from this study were genetically and biochemically similar to those found elsewhere, including well-described human source strains such as WB. We conclude that potentially human-infective Giardia cysts are commonly found in raw surface waters and sewage in Canada, although cyst viability is frequently low. Cryptosporidium oocysts are less common in Canada. An action level of three to five Giardia cysts per 100 liters in treated drinking water is proposed on the basis of the monitoring data from outbreak situations. This action level is lower than that proposed by Haas and Rose (C. N. Haas and J. B. Rose, J. Am. Water Works Assoc. 87(9):81-84, 1995) for Cryptosporidium spp. (10 to 30 oocysts per 100 liters).