We examined whether errorless learning (EL) and learning by modeling (LM) were more advantageous than trial and error learning (TEL) in the acquisition of instrumental activities of daily living (IADL) in Alzheimer's dementia (AD) patients (n = 14). Using a counterbalanced within-subject design, participants performed 3 learning conditions. EL consisted of straightforward prompts before any action, LM focused on the modeling of each step of the tasks and standard TEL without cues was used as a control condition. The participants had to (re)learn 3 IADL. Repeated-measure analyses during learning and follow-up assessments were performed 1 and 3 weeks after learning. The LM and the EL procedures resulted in significantly better learning compared to TEL, with effect sizes (partial eta squared) of 0.42 and 0.35, respectively. This is the first controlled study to show that (re)learning of IADL is possible in patients with AD using an error-reduction approach.