Abstract We reply to the remarks by Ferretti et al. (in this issue) on our paper (De Marco et al., 2014) and demonstrate that: 1. the differences in estimating AOT40 by passive sampling when using the formula proposed by either Mazzali et al. (2002) or De Marco et al. (2014) are not statistically significant; 2. the inaccuracy of the Loibl approach does not strictly depend on the averaging period, as differences between weekly and biweekly calculations are not significant. We thus confirm the well-known uncertainties of AOT40 estimates based on the Loibl function and recommend further testing of the multi-variate statistical model proposed by De Marco et al. (2014). The main advantages of this modelling approach are the direct estimate of AOT40, the applicability at any elevation, and the use of simple geostatistical attributes that can be easily obtained at any location by GIS and existing databases.