Background An association between a nonmedicinal herbal diet and nasopharyngeal carcinoma (NPC) has often been hypothesized but never thoroughly investigated. Methods This study enrolled a total of 2469 patients with incident NPC and 2559 population controls from parts of Guangdong and Guangxi Provinces in southern China between 2010 and 2014. Questionnaire information was collected on the intake of traditional herbal tea and herbal soup as well as the specific herbal plants used in soups and other potentially confounding lifestyle factors. Multivariate logistic regression models were used to estimate odds ratios (ORs) with 95% confidence intervals (CIs) for the NPC risk in association with herbal tea and soup intake. Results Ever consumption of herbal tea was not associated with NPC risk (OR, 1.03; 95% CI, 0.91‐1.17). An inverse association was observed for NPC among ever drinkers of herbal soup (OR, 0.78; 95% CI, 0.67‐0.90) but without any monotonic trend with an increasing frequency or duration of herbal soup consumption. Inverse associations with NPC risk were detected with 9 herbal plants used in herbal soup, including Ziziphus jujuba, Fructus lycii, Codonopsis pilosula, Astragalus membranaceus, Semen coicis, Smilax glabra, Phaseolus calcaratus, Morinda officinalis, and Atractylodes macrocephala (OR range, 0.31‐0.79). Conclusions Consuming herbal soups including specific plants, but not herbal tea, was inversely associated with NPC. If replicated, these results might provide potential for NPC prevention in endemic areas. Consuming herbal soups is inversely associated with nasopharyngeal carcinoma in the endemic region of southern China. Nine herbal types are identified as responsible for this inverse association.