Background & Aims The aim of this study was to characterize the relationship between the broad spectrum of hypothyroidism and NAFLD. Methods A cross-sectional study with 4648 health check-up subjects (2324 cases with hypothyroidism vs. age- and sex-matched controls) was conducted. The subjects were categorized as having either subclinical thyroid-stimulating hormone (TSH) ⩾4.1 mIU/L and normal free thyroixine (T4 ) level (0.7–1.8 ng/dl) or overt hypothyroidism free T4 <0.7 ng/dl. NAFLD was diagnosed on the basis of typical ultrasonographic findings, and alcohol consumption of less than 20 g/day in the absence of other causes of liver disease. Results The mean age of the subjects was 48.6 ± 11.8 years and 62.4% were female. NAFLD was significantly associated with hypothyroidism (30.2% patients vs. 19.5% control, p <0.001). The prevalence of NAFLD and abnormal liver enzyme levels (ALT >33/25 IU/L) increased steadily with increasing grades of hypothyroidism (for NAFLD, subclinical: 29.9% and overt: 36.3%; for abnormal ALT, 20.1% and 25.9%, p <0.001, respectively). Multivariate regression analysis showed that NAFLD was statistically significantly associated with hypothyroidism (odds ratio (OR) 1.38, 95% confidence interval (CI), 1.17–1.62) and the grade of hypothyroidism in a dose-dependent manner (OR 1.36, 95% CI, 1.16–1.61 in subclinical hypothyroidism and OR 1.71, 95% CI, 1.10–2.66 in overt hypothyroidism). Conclusions Subclinical hypothyroidism, even in the range of upper normal TSH levels, was found to be related to NAFLD in a dose-dependent manner. Hypothyroidism is closely associated with NAFLD independently of known metabolic risk factors, confirming a relevant clinical relationship between these two diseases.