The possibility that the high frequency of treatment failures in Indian kala-azar might be due to infection with antimony-resistant strains of Leishmania donovani has not been experimentally addressed. L. donovani isolates were obtained from splenic aspiration smears of 24 patients in Bihar, India, who either did not respond (15) or did respond (9) to 1 or more full courses of treatment with sodium antimony gluconate (SAG). A strong correlation (P<.001) between clinical response and SAG sensitivity in vitro was observed only when strains were assayed as intracellular amastigotes: responsive isolates ED50=2.4+/-2.6, ED90=6.4+/-7.8 microgram SAG/mL; unresponsive isolates ED50=7.4+/-3.7 microgram SAG/mL, ED90=29.1+/-11.1 SAG/mL. No correlation with clinical response was found by use of extracellular promastigotes (ED50=48+/-22 vs. 52+/-29 microgram/mL). The emergence of antimony-resistant L. donovani strains appears to be a cause of treatment failures in India.