This paper examines the hypothesis of GDP per capita convergence in the northeastern states, the period 1995-2009, with a focus on human capital. The reason to analyze the Northeast only and not all regions of Brazil is to retain the condition of homogeneity, necessary to analyze convergence issues. Using theories and models of economic growth, we tested the hypothesis of convergence ß- (absolute and conditional) convergence, and -o- during the period analyzed. Using Stata software version 12.1, the coefficients were estimated using panel data with fixed effects, whose main advantage is in capturing behavior changes of the indicators analyzed, obtaining more data, informative and efficient, which would not be possible with only data cross section or time series. The results pointed to the existence of convergence-o-, ß- and ß--absolute-conditional, obtaining, in the case of absolute convergence, growth rates well above those found in the empirical literature. Under the use of proxies of human capital, the conditional convergence proved apparently differentiated between samples considered, revealing the persistence of differences interstate. / O presente trabalho analisa a hipotese da convergencia do PIB per capita nos estados nordestinos, no periodo de 1995-2009, com foco no capital humano. A necessidade de analisar apenas a Regiao Nordeste e nao todas as Regioes do Brasil prendeu-se a satisfazer a condicao de homogeneidade, necessaria para tratar a questao da convergencia. Tendo como aparato as teorias e modelos de crescimento economico, foram testadas as hipoteses de convergencia ß (absoluta e condicional) e convergencia-o- durante o periodo analisado. Utilizando o Software Stata versao 12.1, os coeficientes foram estimados atraves de dados em painel com efeitos fixos, que tem como principal vantagem captar as mudancas no comportamento dos indicadores analisados, obtendo dados mais, informativos e eficientes, o que nao seria possivel apenas com dados cross section ou com series temporais. Os resultados apontaram para a existencia de convergencia-o-, ß-absoluta e ß-condicional, obtendo, no caso da convergencia absoluta, taxas de crescimento bem superiores as encontradas na literatura empirica. Sob a utilizacao de proxies do capital humano, a convergencia condicional mostrou-se aparentemente diferenciada entre as amostras consideradas, revelando a persistencia das diferencas interestaduais.