This talk will focus on UV nadir measurements of SO2 at Venus cloud top level obtained with SPICAV/Venus Express from 2006 to 2010, using an expanded data set and updated version of Marcq et al. [2008] radiative transfer algorithm. Provisional results show a more complicated picture than thought based on the first 300 Venus Express orbits [Marcq et al., 2011; Belyaev et al., 2008]. Short-term variability as well as maximal values seem on par with 1980s UVS/Pioneer Venus data [Esposito et al., 1988], confirming the halt of the negative trend in SO2 mean abundance observed in the 1980s and 1990s. Latitudinal gradient is usually negative, but positive in a handful of orbits, reminiscent of the situation in the 1980s [Zasova et al., 1993; Na et al., 1994]. This may point to frequent changes in the relative intensity of the processes ruling SO2 abundance at cloud top level, leading to at least two distinct regimes for SO2 with opposite latitudinal gradients. A positive correlation between the mean UV brightness (proxy for unknown UV absorber) and SO2 mixing ratio has also been evidenced. Naive transport modeling as well as much more sophisticated chemical modeling [Krasnopolsky, 2012] seem able to reproduce much of the observed spatial and temporal variability of SO2, indicating at least a qualitative understanding of the physical and chemical process at work. All these results will be compared when relevant to the band ratio SO2 retrievals from Pioneer Venus era [Esposito et al., 1988] in order to investigate possible changes on a decadal timescale in Venus' lower mesosphere.