Prediction of survival in frontotemporal lobar degeneration (FTLD) is guesswork. The aim of the present study was to evaluate whether SPECT scan may be useful to predict prognosis of long term survival in FTLD patients. A cohort of 125 patients with FTLD who underwent brain SPECT scan at the time of enrollment and who were further followed up for at least one year were considered. In each subject, volume of interests (VOIs) covering frontotemporal and parietal regions, bilaterally, were drawn. Principal component analysis (PCA) was applied on VOIs, and a Cox regression model was carried out to find out best predictors of survival. A two-pattern PCA solution was chosen, explaining more than 70% of variance, and "frontal" PC1 and "temporal" PC2 components were identified. The frontal PC1 was associated with higher rate of faster progression (HR = 2.06, 95% CI = 1.23-3.44, p = 0.006 for univariate model, and HR = 1.85, 95% CI = 1.04-3.28, p = 0.03 for multivariate model). In particular, right orbitofrontal cortex showed the higher loadings in PC1; the worse the scores of this region the shorter the survival was reported. We suggest that SPECT imaging, beyond a helpful tool in diagnostic assessment, may be an easily and accessible marker of disease outcome in FTLD. Further studies considering structural neuroimaging are warranted.