Background: Only a subgroup of subjects with probable AD shows cognitive improvement after treatment with cholinesterase inhibitors. Objectives: To investigate whether atrophy of the medial temporal lobe and the apolipoprotein E (APOE) genotype could predict cognitive improvement in subjects with probable AD treated with rivastigmine. Methods: 121 subjects with mild to moderate probable AD were treated for 26 weeks with escalating doses of rivastigmine. Outcome measures were change on the MMSE and the ADAS-Cog between baseline and follow-up and treatment response defined as at least 2 points improvement on the MMSE or at least 4 points improvement on the ADAS-Cog. The study was an open-label multi-centre study in the Netherlands. Results: Subjects with medial temporal lobe atrophy (MTLA) tended to show more decline on the MMSE after correction for age, sex, education, baseline cognitive score, and dosage at week 26 compared with subjects without MTLA (p = 0.08). A significant interaction between MTLA and dosage at week 26 existed for change on the MMSE and ADAS-Cog: subjects with MTLA showed more cognitive decline than subjects without MTLA only in the group of patients who received a low dosage at week 26. MTLA was not associated with treatment response. The APOE genotype was not associated with change on the MMSE or ADAS-Cog or with treatment response. Conclusions: MTLA and the APOE genotype are not clinically useful predictors of cognitive response in subjects with probable AD who are treated with rivastigmine.