We use the final data of the VIMOS Public Extragalactic Redshift Survey ( VIPERS ) to investigate the effect of the environment on the evolution of galaxies between z = 0.5 and z = 0.9 . We characterise local environment in terms of the density contrast smoothed over a cylindrical kernel , the scale of which is defined by the distance to the fifth nearest neighbour . This is performed by using a volume-limited sub-sample of galaxies complete up to z = 0.9 , but allows us to attach a value of local density to all galaxies in the full VIPERS magnitude-limited sample to i < 22.5 . We use this information to estimate how the distribution of galaxy stellar masses depends on environment . More massive galaxies tend to reside in higher-density environments over the full redshift range explored . Defining star-forming and passive galaxies through their ( NUV - r ) vs ( r - K ) colours , we then quantify the fraction of star-forming over passive galaxies , f _ { ap } , as a function of environment at fixed stellar mass . f _ { ap } is higher in low-density regions for galaxies with masses ranging from \log ( \mathcal { M } / \mathcal { M } _ { \odot } ) = 10.38 ( the lowest value explored ) to at least \log ( \mathcal { M } / \mathcal { M } _ { \odot } ) \sim 11.3 , although with decreasing significance going from lower to higher masses . This is the first time that environmental effects on high-mass galaxies are clearly detected at redshifts as high as z \sim 0.9 . We compared these results to VIPERS-like galaxy mock catalogues based on a widely used galaxy formation model . The model correctly reproduces f _ { ap } in low-density environments , but underpredicts it at high densities . The discrepancy is particularly strong for the lowest-mass bins . We find that this discrepancy is driven by an excess of low-mass passive satellite galaxies in the model . In high-density regions , we obtain a better ( although not perfect ) agreement of the model f _ { ap } with observations by studying the accretion history of these model galaxies ( that is , the times when they become satellites ) , by assuming either that a non-negligible fraction of satellites is destroyed , or that their quenching timescale is longer than \sim 2 Gyr .