We aimed to assess the independent effect of clinical spectrum, bacterial inoculum size and physician characteristics on the sensitivity of a rapid antigen detection test (RADT) for group A streptococcus (GAS) in children. Double throat swabs were collected from 1,482 children with pharyngitis and 294 asymptomatic children in a French prospective, office-based, multicenter (n = 17) study, from October 2009 to May 2011. Patient- and physician-level factors potentially affecting RADT sensitivity were studied by univariate and multivariate multilevel analysis, with laboratory throat culture as the reference test. In children with pharyngitis and asymptomatic children, the prevalence of GAS was 38 % (95 % confidence interval 36-41 %) and 11 % (7-14 %), respectively. Overall, RADT sensitivity was 87 % (84-90 %). On stratified and multivariate multilevel analysis, RADT sensitivity was higher for children with pharyngitis than asymptomatic children (89 % vs. 41 %), children <9 than ≥ 9 years old (88 % vs. 79 %) and those with heavy than light inoculum (94 % vs. 53 %). RADT sensitivity was influenced by the physician performing the test (range 56-96 %, p = 0.01) and was higher for physicians with hospital-based clinical activity in addition to office-based practice (adjusted odds ratio 3.4 [95 % confidence interval 1.9-6.3], p < 0.001); inter-physician variations in RADT sensitivity were largely explained by this variable (proportional change in variance >99 %). The sensitivity of the RADT is independently affected by patient- and physician-level factors. Physicians who base their diagnosis of GAS pharyngitis on the results of a RADT alone should consider diagnostic accuracy monitoring and adequate training when needed.