Purpose Depending on the calibration of their cognitive abilities, some older drivers (ODs) might stop driving prematurely (under-estimators, UEs) and others could expose themselves to risky situations (over-estimators, OEs). The aim of the study was to compare the effectiveness of two cognitive training (CT) programs intended for ODs presenting a cognitive calibration bias. We hypothesized that CT with feedback on performance can help ODs to correctly calibrate their abilities and consequently adapt their driving behavior. Method One hundred and six ODs (≥70 years) were assigned to two CT groups (with or without a driving simulator experience, DS). These interventions lasted about 36 h and were distributed over a 3-month period. ODs completed objective and subjective cognitive evaluations and an on-road driving evaluation before and after training. Results The first results on 67 participants (40 from the CT group, and 27 from the CT + DS group) showed an improvement of their visual processing speed, their divided attention and their selective attention after training. Participants from both groups also had an improved TRIP tactical sub-score (Test Ride for Investigating Practical fitness to drive), indicating a better driving behavioral adaptation. Finally, although both training programs seemed to be equally effective in correcting cognitive calibration bias, the results indicated that 21 UEs and 10 OEs were well calibrated and thus correctly self-assessed their cognitive abilities after training. Conclusion Both CT programs (with or without DS experience) seem to improve the visual attention of ODs. UEs appeared to be more susceptible than OEs to this training and were better calibrated after it.