To detect fire-induced variability in vegetational dynamics, a time series 1998 to 2003 of Normalized Difference Vegetation Index (NDVI) from SPOT-VEGETATION sensor was analyzed for burned and unburned test sites located in the Italian Peninsula. The statistical analysis was performed by using the detrended fluctuation analysis (DFA), which permits the detection of persistent properties in nonstationary signal fluctuations. A comparison of results obtained for sites in "healthy condition " (fire un-unaffected) and in "ill condition" (fire affected) showed that fires contribute in increasing the persistence of time dynamics of vegetation. This result indicates that fires drive unstable patterns in vegetational dynamics of burned areas. The adopted methodology could be fruitfully applied to investigate other types of vegetational stresses.