This study assesses long-run real per capita output convergence among selected Latin American countries. The empirical investigation, however, is based on an alternative approach. Strong convergence is determined on the basis of the first largest principal component, based on income differences with respect to a chosen base country, being stationary. The qualitative outcome of the test is invariant to the choice of base country and, compared to alternative multivariate tests for long-run convergence, this methodology places less demands on limited data sets. Using annual data for the period 1960-2000, strong convergence is confirmed for the Central American Common Market. However, an amended version of the test confirms weaker long-run convergence in the case of the Latin American Integration Association countries.