Objectives: Metagenomic next-generation sequencing (mNGS) has been applied more and more widely for the diagnosis of infectious diseases, but its performance in the diagnosis of cryptococcal meningitis (CM) remains unclear.Methods: Cerebrospinal fluid (CSF) samples from 197 HIV-negative patients with suspected central nervous system infections were tested simultaneously by mNGS and routine methods [India ink staining, fungal culture, or cryptococcal antigen (CrAg) tests]. The performance of mNGS was evaluated.Results: Of the 197 enrolled cases, 46 (23.4%) cases were finally diagnosed with CM, including 43 (93.5%) Cryptococcus neoformans infections and 3 (6.5%) Cryptococcus gattii infections. The sensitivity, specificity, positive predictive value, negative predictive value, and concordance rate of mNGS were 93.5% [95% confidence interval (CI) at 86.4%~100.0%], 96.0% (95% CI at 92.9%~99.1%), 87.8%, 98.0%, and 95.4%, respectively. Comparing to the conventional diagnostic methods, the sensitivity and concordance rate of mNGS were slightly lower than those of CrAg tests (97.4%) but higher than those of India ink (63.0%) and culture (76.7%). Besides, mNGS had a sensitivity of 100.0% against culture. It should be noted that mNGS could identify Cryptococcus at species level; C. gattii of the 3 cases was only distinguished by mNGS.Conclusions: CSF mNGS can be considered as a supplementary test to diagnose CM and directly distinguish C. gattii from C. neoformans in clinical specimens.