Aims To compare selective and non-selective TACE techniques in the treatment of HCC with a special emphasis on clinical and liver tolerance, tumour response and survival. Methods 184 patients with advanced HCC were retrospectively included. Three different TACE techniques were compared: non selective lipiodol-chemotherapy + non selective embolisation (TACE-technique group 1), non selective lipiodol-chemotherapy + selective embolisation (group 2), and selective lipiodol-chemotherapy + selective embolisation (group 3). Results In multivariate analysis TACE-technique group is an independently significant prognostic factor for poor clinical tolerance, poor liver tolerance and tumour response. The rate of patients with poor clinical tolerance was lower in group 3 (27.0%) than in groups 1 (64.1%, p < 10−3) or 2 (66.7%, p < 10−3). The rate of patients with poor liver tolerance was higher in group 2 (34.0%) than in groups 1 (17.6%, p = 0.050) or 3 (6.9%, p = 0.011). The rate of patients with tumour response was higher when embolisation was selective versus non-selective, i.e., group 2 + 3 (78.7%) versus group 1 (62.5%, p = 0.054). Overall survival was not significantly different between the three groups (p = 0.383). Conclusion Both selective techniques resulted in better tumour response. As for improving tolerance, our study suggests that the main technical factor is the use of selective lipiodol-chemotherapy injection.