Background We aimed to determine if there was a higher incidence of small intestinal bacterial overgrowth (SIBO) in non‐alcoholic fatty liver disease (NAFLD) than in patients without NAFLD. Moreover, we assessed whether patients with significant fibrosis (SF) had a higher incidence of SIBO compared with patients with non‐significant or no liver fibrosis. Methods NAFLD was diagnosed in 117 patients by using Fibroscan with a controlled attenuation parameter (CAP) as well as liver biopsy (LB). SIBO was defined by esophagogastroduodenoscopy with an aspiration of the descending duodenum. Results Patients with non‐alcoholic steatohepatitis (NASH) and those with SF on LB had a significantly higher incidence of SIBO than patients without NASH and those without SF, respectively (P < .05). According to histological characteristics, there was a higher proportion of patients in the SIBO group with higher steatosis and fibrosis grade, lobular and portal inflammation, and ballooning grade (P < .001). In multivariate analysis, significant predictors associated with SF and NASH were type 2 diabetes mellitus (T2DM) and SIBO. Moreover, in multivariate analysis, significant predictors that were independently associated with SIBO were T2DM, fibrosis stage and ballooning grade (OR 8.80 (2.07‐37.37), 2.50 (1.16‐5.37) and 27.6 (6.41‐119), respectively). The most commonly isolated were gram‐negative bacteria, predominantly Escherichia coli and Klebsiella pneumoniae. Conclusion In this relatively large population of patients, we used a gold standard for both SIBO (quantitative culture of duodenum's descending part aspirate) and NAFLD (LB), and we demonstrated that NASH patients and those with SF had a higher incidence of SIBO. Moreover, significant predictors independently associated with SIBO were T2DM, fibrosis stage and ballooning grade. Although TE is a well‐investigated method for steatosis and fibrosis detection, in our study, independent predictors of SIBO were histological characteristics of NAFLD, while elastographic parameters did not reach statistical significance.