The STELLA-LONG TERM prospective post-marketing surveillance study assessed ipragliflozin in Japanese patients with type 2 diabetes mellitus (T2DM). This subgroup analysis of patients with liver impairment used the final 3-year results. Data on patients, adverse drug reactions (ADRs), and changes in glycemic parameters and liver enzymes (aspartate aminotransferase [AST], alanine aminotransferase [ALT], gamma-glutamyl transpeptidase [γ-GTP] and alkaline phosphatase [ALP]) were collected, and the fatty liver index (FLI) was calculated. In the effectiveness analysis (n = 8,763), baseline liver function was normal in 2,605 patients (ALT <31/<21 U/L [men/women]) and abnormal in 3,277 (ALT ≥31/≥21 U/L). The abnormal liver function group had higher mean body weight and BMI than the normal liver function group (p < 0.001). In the safety analysis (n = 11,051), urinary tract infections, genital infections and hepatic disorders were more common in the abnormal than normal liver function group (2.25% vs. 1.07%; 1.78% vs. 1.14% and 1.85% vs. 1.01%). In the abnormal liver function group, there were significant (p < 0.001) decreases from baseline at 36 months in AST and ALT (from 38.8 and 53.7 U/L to 29.3 and 37.7 U/L, respectively), γ-GTP (from 75.4 to 51.7 U/L) and ALP (from 254.8 to 234.5 U/L), which were greater than in the normal liver function group. FLI reductions at 36 months were significant (p < 0.001) in subgroups with baseline FLI of ≥30 or ≥60. In conclusion, ipragliflozin improved liver function over 3 years in patients with impaired liver function, although ADRs occurred more frequently than in the normal liver function group.