In the past decade, convergence of regional per capita income has been a frequent object of regional studies- In the case of Indonesia, research on this issue has been exploited for many years- However, differences in methods of analysis often yield different results. Further more, prior studies do not elaborate the advantages of each method. This paper will focus on this issue with stressing on searching of the appropriate methods-Rather a-convergence, we use Theil index to analyze regional disparity and found that regional convergence in Indonesia does not appear to be a simple monotonic process, but seems to vary over time -and hence requires explanation. From here, we suggest that the use of long-term interval in regression of &? convergence analysis will destroy the relationship among variables. The use of least square methods only suitable for absolute convergence analysis. We conclude that the most appropriate methods for conditional ^convergence are instrumental methods. As an alternative one, we suggest panel-data methods.We found that the convergence rate in Indonesia is slow at 2,59 percent per year over 1975-2000. With conditional convergence analysis, we found that the difference on growth rate can be explained systematically by a set of explanatory variables. The inclusion of these variables makes the convergence rate become faster which is 3.97 percent per year. Most of this variables are under controlled by government. In summary, we conclude that government policies have significant influence for rapid and sustainable regional economic growth.