Summary Background The mcr-1 gene confers transferable colistin resistance. mcr-1 -positive Enterobacteriaceae (MCRPE) have attracted substantial medical, media, and political attention; however, so far studies have not addressed their clinical impact. Herein, we report the prevalence of MCRPE in human infections and carriage, clinical associations of mcr-1 -positive Escherichia coli (MCRPEC) infection, and risk factors for MCRPEC carriage. Methods We undertook this study at two hospitals in Zhejiang and Guangdong, China. We did a retrospective cross-sectional assessment of prevalence of MCRPE infection from isolates of Gram-negative bacteria collected at the hospitals from 2007 to 2015 (prevalence study). We did a retrospective case-control study of risk factors for infection and mortality after infection, using all MCRPEC from infection isolates and a random sample of mcr-1 -negative E coli infections from the retrospective collection between 2012 and 2015 (infection study). We also did a prospective case-control study to assess risk factors for carriage of MCRPEC in rectal swabs from inpatients with MCRPEC and mcr-1 negative at the hospitals and collected between May and December, 2015, compared with mcr-1 -negative isolates from rectal swabs of inpatients (colonisation study). Strains were analysed for antibiotic resistance, plasmid typing, and transfer analysis, and strain relatedness. Findings We identified 21 621 non-duplicate isolates of Enterobacteriaceae, Acinetobacter spp, and Pseudomonas aeruginosa from 18 698 inpatients and 2923 healthy volunteers. Of 17 498 isolates associated with infection, mcr-1 was detected in 76 (1%) of 5332 E coli isolates, 13 (<1%) of 348 Klebsiella pneumoniae , one (<1%) of 890 Enterobacter cloacae , and one (1%) of 162 Enterobacter aerogenes . For the infection study, we included 76 mcr-1 -positive clinical E coli isolates and 508 mcr-1 -negative isolates. Overall, MCRPEC infection was associated with male sex (209 41% vs 47 63%, adjusted p=0·011), immunosuppression (30 6% vs 11 15%, adjusted p=0·011), and antibiotic use, particularly carbapenems (45 9% vs 18 24%, adjusted p=0·002) and fluoroquinolones (95 19% vs 23 30%, adjusted p=0·017), before hospital admission. For the colonisation study, we screened 2923 rectal swabs from healthy volunteers, of which 19 were MCRPEC, and 1200 rectal swabs from patients, of which 35 were MCRPEC. Antibiotic use before hospital admission (p<0·0001) was associated with MCRPEC carriage in 35 patients compared with 378 patients with mcr-1-negative E coli colonisation, whereas living next to a farm was associated with mcr-1-negative E coli colonisation (p=0·03, univariate test). mcr-1 could be transferred between bacteria at high frequencies (10−1 to 10−3 ), and plasmid types and MCRPEC multi-locus sequence types (MLSTs) were more variable in Guangdong than in Zhejiang and included the human pathogen ST131. MCRPEC also included 17 unreported ST clades. Interpretation In 2017, colistin will be formally banned from animal feeds in China and switched to human therapy. Infection with MRCPEC is associated with sex, immunosuppression, and previous antibiotic exposure, while colonisation is also associated with antibiotic exposure. MLST and plasmid analysis shows that MCRPEC are diversely spread throughout China and pervasive in Chinese communities. Funding National Key Basic Research Program of China, National Natural Science Foundation of China/Zhejiang, National Key Research and Development Program, and MRC, UK.