This paper examines the relationship among human capital, total factor productivity growth and convergence using international panel data on macroeconomic indicators and educational attainment. It explicitly allows for the heterogeneity in technology growth across countries by use of the stochastic Solow model and dynamic panel estimation techniques. Thus the convergence in this paper is toward the steady state of the individual country, not toward the common steady state for all nations in the data set as in the existing literature. It reports the findings as follows. First, heterogeneity in technology growth across countries, which has been typically assumed to be nonexistent in the growth literature, is found to be prominent. Second, the average estimates of convergence rates are between 27% and 32% per annum, which are much higher than those reported in the researches using cross-section data. Third, the model with human capital proxied by various educational attainment measures gives higher convergence rate estimates than the model without it. Fourth, the average estimates for technology growth are higher in the model with human capital than in the model without it. Fifth, it estimates the structural equation in which human capital levels can influence total factor productivity growth by domestic endogenous innovation and technological catch-up effects. It finds that the catch-up effect is stronger than the domestic innovation effect.