This paper aims at investigating the trade and convergence in per capita income link by applying the threshold methodology developed by Hansen [2000] to standard growth regressions in order to capture a non-linear effect of trade on growth. We divide our ten trade measures into two broad categories: trade intensity ratios and measure of trade restrictions. Trade indicators are the most important variables to explain the clustering of countries. Amongst the ten trade measures applied, there is a difference between the two broad measures. The null of linearity has always been rejected in favour of the threshold regression with the trade intensity ratios only. A second threshold may be found with either the initial per capita income or the population growth rate. Finally, we have selected three different regimes. The relationship between growth and its determinants is diverse for each regime. Countries belonging to regimes 1 and 3 are diverging. The conditional convergence hypothesis is only accepted in the second regime. This finding implies that convergence in per capita income is not uniform across countries: there is a catching-up movement mainly between developed countries and a few developing countries. For the vast majority of developing countries, divergence in per capita income seems to be the norm rather than the exception. The correlation between measures of trade restrictions and growth is different across regimes. The correlation is non significant in the first and third regime. The positive effect of a decrease in tariffs on growth will depend on the level of development. For the majority of the developing countries included in our sample a decrease in tariffs will have no effect on growth.