ABSTRACTBackgroundAdults with Down syndrome (DS) have an elevated risk and early age of onset for Alzheimer's disease (AD). To support upcoming clinical AD trials, there is a critical need to establish cognitive outcome measures that can be used to capture intervention effects. One measure that has successfully been used to detect AD‐related cognitive decline in the DS population is a measure of episodic memory, the modified Cued Recall Test (mCRT). Demonstrated utility of the mCRT warrants further investigation into comparisons between the A and B versions, free versus cued recall and changes in performance over time to better understand sensitivity for tracking memory decline over time based on age and AD clinical status.MethodParticipants were 272 adults with DS aged 25–81 (mean age = 43.12 years, SD = 9.79). Study procedures were completed at three cycles of data collection: baseline, 16‐month follow‐up and 32‐month follow‐up. Participants were enrolled in the Alzheimer Biomarker Consortium–Down Syndrome longitudinal study and completed the mCRT as part of a multiday evaluation. Comparisons were made between the A and B versions of the mCRT in recall and intrusion scores. Participants' ratio of free relative to cued recall was also examined at baseline and longitudinally. Participant performance was compared by age group, clinical AD status (cognitively stable [CS], mild cognitive impairment [MCI] or AD dementia) and premorbid level of intellectual disability (ID).ResultsVersion differences were identified, with the most salient differences in the moderate and severe/profound ID groups. The mCRT free recall declined with age in CS participants. Free and cued recall scores were lower in those with MCI and AD dementia, with the exception of the mild ID MCI group, whose cued recall scores were not significantly different from the CS group. Decline across 32 months (mCRT total score decline of 1.29 points/year) was observed for CS participants beginning at ≥ 50 years old, with more pronounced declines in adults with DS with an MCI or AD dementia diagnosis (3.36 and 4.20 points/year, respectively).ConclusionCharacterising test version differences and participant free versus cued recall performance on the mCRT is important for understanding performance under testing conditions and to maximise the sensitivity of clinical interventions to capture meaningful effects. Our findings suggest that clinical AD trials for DS should be cautious about using both versions of the mCRT. Examining the profile of free relative to cued recall may enhance sensitivity for detecting treatment benefits for adults with DS across the range of premorbid ID levels.