Objective. To evaluate validity of the Mini-Mental State Examination (MMSE) that supports the interpretability in older adults. Methods. Secondary data study administered in 4.034 older adults in Antioquia, Colombia, 2012. Data were analyzed from Rasch models according to Wolfe and Smith guidelines. Results. Attention and calculation items showed local dependency. They produce a second relevant measure affecting the unidimensional type of the test. The reliability of the items was of 0.80, the measure explained 45.2% of the raw score variance. Conclusion. Results suggest differentiating attention and calculation abilities from the other cognitive processes considered by the test. MMSE shows generalization validity limitations that may affect transcultural comparisons.