AbstractObjectivesTo investigate the course of activities of daily living (IADL) functioning and possible predictors of performance changes in healthy older adults conducting either a General Cognitive Training (GCT) or a Reasoning Cognitive Training (ReaCT) or no training (control group, CG) over a period of 6 weeks, 3 months, and 6 months.Setting and participantsAn online, home‐based GCT and ReaCT including n = 2913 healthy participants (GCT: n = 1096; ReaCT: n = 1022; CG: n = 794) aged 60 years and older.MethodsMultilevel analysis were calculated to explore the nature of our outcome variables of IADL part A (independence) and part B (difficulty of tasks), and to detect possible predictors for participants’ performance on IADL after CT.ResultsThe random slopes models fitted better for the outcomes IADL Part B in the GCT group (χ²(2) = 18.78, p < .01), and both IADL Part A and Part B in the ReaCT group (χ²(2) = 28.57, p < .01; χ²(2) = 63.38, p < .01, respectively), indicating different changes over time for individual participants. Female sex was a significant predictor of IADL change in the ReaCT group, showing that females benefited most in both IADL scores (IADL A: 0.01, p < .01; IADL B: 0.004, p < .01). No other significant predictors for IADL changes were identified.Conclusion and implicationThe particular effectiveness in women is of clinical relevance, as IADL is typically more impaired in women than in men in advanced age. Following a personalized medicine approach, identifying predictors of non‐pharmacological intervention success is of utmost importance.