Alzheimer's disease (AD) is associated with excess whole brain volume loss, and progressive cognitive impairment. We aimed to study the extent to which these two potential biomarkers of AD progression are correlated. Forty-six patients with sporadic AD were tested with a neuropsychometric battery including test of verbal and visual memory, vocabulary, arithmetic, naming, visuoperceptual skills and reasoning at two time-points, approximately 1 year apart; annualised rates of change for each test were calculated. Each subject also attended for up to twelve T1-weighted volumetric MRI scans at fixed intervals over a 2-year period. For each individual all possible scan-pairs were positionally registered, and whole brain atrophy rates were calculated using the brain boundary shift integral. Linear mixed models were used to investigate associations between atrophy rate and coincident change in each neuropsychometric score. Each model estimated the effect of a unit change in score, plus the additional effect of a fall to floor, after adjusting for baseline levels. 467 MRI scans were performed, permitting 2199 individual measures of change to be made. The model-derived mean atrophy rate was 2.23% per year with a between-subject SD of 0.99% per year. Increasing atrophy rate was significantly associated with rate of change in a number of non-memory based neuropsychological scores, with the strongest association seen with longitudinal change in matrix reasoning (p=0.004). These results provide further evidence that cerebral atrophy is a clinically relevant marker of AD progression. This methodology whereby data from patients falling to floor on a given test may be included and accounted for, rather than discarded, may find broader application in clinical studies incorporating neuropsychometric outcomes.