During the past few years, the incidence of invasive group A Streptococcus (GAS) infection has been increasing. However, there are presently no clear recommendations regarding antibiotic prophylaxis for close contacts of index patients. The aims of this study were 1) to determine the prevalence of carriage of the same GAS strain as the patient's among contacts of patients with invasive infections and 2) to assess the importance of exposure duration. From March 1995 to March 1996, the authors prospectively included in the study all patients with invasive GAS infection, as defined by the Working Group on Severe Streptococcal Infections, who came to Hôpital Maisonneuve-Rosemont in Montreal, Quebec, Canada. An epidemiologic investigation was systematically carried out for each index case. Contacts were divided into two groups: those who had spent 24 hours or more with the index patient during the week preceding the beginning of his or her illness and those who had spent 12-24 hours with the index patient during that week. Strains of GAS were examined by serotyping (proteins M and T and the presence or absence of the serum opacity factor) and by characterization of streptococcal pyrogenic exotoxins (exotoxins A, B, and C). One hundred and two contacts of 17 index cases with invasive GAS infection were systematically screened. Contacts were considered positive if they carried the same strain of the bacterium and the same streptococcal pyrogenic exotoxin as the index case. Among the contacts who had spent at least 24 hours per week with their respective index cases, 13 out of 48 (27%) were found to be harboring the same serotype of GAS as the index patient (95% confidence interval 14.5-39.5). By comparison, only one of the 54 contacts in the 12- to 24-hour group (1.8%) was found to be carrying the same strain of the bacterium (95% confidence interval 0-5.3). This difference between the two groups was statistically significant (p < 0.001). The median age of the positive carriers (10 years) was significantly lower than the median age of the noncarriers (39 years) (p ≤ 0.0005). This study showed that close contacts who had spent 12-24 hours with the index patient were rarely colonized with GAS. If antibiotic prophylaxis against GAS is recommended, it should probably target contacts who spent at least 24 hours with an infected patient during the week preceding illness onset.