Background: Radiomics is a valuable tool for predicting hematoma expansion (HE) but has not been used for small intracerebral hemorrhage (ICH). We hypothesized that not all small hematomas are benign and that radiomics could predict HE and short-term outcomes in small hematomas.Methods: We analyzed 313 patients with small ICH who underwent baseline noncontrast CT within 6 h of symptom onset between September 2013 and February 2019. Small ICH was defined as baseline hematoma volume <10 mL. A radiomic score (R-score) was developed in a training (n=218) and validated in a test cohort (n=95). Poor outcome was defined as a Glasgow Outcome Scale score ≤3. The relationship of the R-score with HE and outcomes was investigated using univariate and multivariate analyses. Predictive performance was assessed by the area under the receiver operating characteristic (ROC) curve (AUC).Results: R-score was an independent predictor of HE in the training (odds ratio [OR]: 2.557; 95% CI, 1.455–4.492) and test cohorts (OR, 3.985; 95% CI, 1.051–14.453). In the 3–10 mL subgroup, but not in the <3 ml subgroup, the R-score was independently associated with HE (OR, 4.293; 95% CI, 2.095–8.796) and poor outcome (OR, 1.297; 95%CI, 1.004–1.674) after adjusting for confounders. The R-score achieved good discrimination ability for HE in the training and test cohorts and the 3–10 mL subgroup (AUCs 0.728, 0.717, and 0.740, respectively).Conclusions: Radiomics provides an objective and effective approach for discriminating between benign and malignant course in patients with small ICH, particularly 3–10 mL hematomas.