APA guidelines for the evaluation of age-related cognitive decline and dementia emphasize the need for baseline (premorbid) data against which current performance can be compared. As this information rarely exists, clinicians must rely on instruments especially designed for estimation of premorbid abilities. No such instrument was available in Portugal until the development of the TeLPI, an irregular words oral reading test. This study aims to examine TeLPI's validity as a measure of premorbid ability in the spectrum of aging cognitive decline, from mild cognitive impairment (MCI) to moderate Alzheimer disease (AD), by the analysis of its stability in normal versus impaired samples. A total of 104 patients, classified into 2 clinical groups, MCI (n=53) and probable mild to moderate AD (n=51), were compared with a group of cognitively healthy controls (C_MCI: n=53; C_AD: n=51) and matched for sex, age, education, and residence. As expected, the Mini-Mental State Examination and Montreal Cognitive Assessment results were significantly different between the groups (AD<MCI<controls), reflecting the severity of cognitive impairment. TeLPI median scores of controls, MCI, and probable AD patients were comparable after correcting for years of education, revealing no significant effect of cognitive impairment on TeLPI performance, and suggesting its validity for estimating premorbid intelligence in subjects with cognitive decline and dementia.