Bemisia tabaci can harbor both primary and secondary endosymbionts, and the specific endosymbionts can differ among different B. tabaci biotypes. This study determined (1) the prevalence of the primary endosymbiont Portiera aleyrodidarum and secondary endosymbionts Arsenophonus and Wolbachia in two invasive biotypes (B and Q) and one indigenous biotype (Cv) in China and (2) the in vivo effect of three antibiotics (tetracycline, ampicillin trihydrate, and rifampicin) against the endosymbionts; if an antibiotic substantially inhibits an endosymbiont, it could be used to determine the effect of that endosymbiont on B. tabaci. P. aleyrodidarum and Wolbachia were detected in all the three biotypes, while Arsenophonus was found only in the Q and Cv biotypes. P. aleyrodidarum was found in all tested individuals of the three biotypes. Infection rates of Wolbachia in the B, Cv, and Q biotypes were 58, 68, and 48%, respectively. The infection rate of Arsenophonus was 44% in the Q biotype but only 22% in the Cv biotype. The antibiotics failed to eliminate P. aleyrodidarum from any individual of the B, Cv, and Q biotypes but eliminated the secondary endosymbionts, Arsenophonus and Wolbachia, from 50 to 80% of the adult B. tabaci. The effect of the antibiotics depended on the species of endosymbiont, the antibiotic, the B. tabaci biotype, and various interactions between these factors. When used against Arsenophonus, the efficiency of rifampicin was better than ampicillin and tetracycline, regardless of B. tabaci biotype. When inactivating Wolbachia in Cv and Q biotypes, the efficiency tetracycline was better than ampicillin and rifampicin, and while the efficiency of tetracycline was better than rifampicin and ampicillin when they were used against Wolbachia in B biotype.