This paper reviews the empirical literature using a gravity model to assess the impact on trade of Preferential Trade Agreements (PTAs) and provides a critical analysis of the choices made to carry out the estimations. With respect to the effectiveness of PTAs, the main result that emerges from the review is that PTAs tend to foster trade between the countries involved. However, two empirical issues appear relevant. First, the use of dummy variables to proxy PTAs does not adequately describe the preferences granted and can be misleading. Indeed, the use of dummies does not allow to discern among the different preferential trade policy instruments used as well as to measure the level of the trade preferences granted. As a result, dummies do not allow to capture the specific effect of the preferences on bilateral trade. Second, the econometric methods used do not always address in a satisfactory way the potential sources of bias in the estimations, such as unobserved heterogeneity, endogeneity of some regressors and zero-trade flows. Thus, with respect to the empirical issues, the main conclusion which can be drawn from this survey is that the proposed estimates so far of the effects of PTAs using gravity models tend to be biased and, as a result, are not fully reliable.