The paper estimates Korean gini coefficients of non-agricultural householdincome distribution for 16 provincial-level administrative regions andthe nation by using public use micro data of National Survey of HouseholdIncome and Expenditures in 1996 and 2000, and tests hypotheses aboutchanges of gini coefficients between the two time periods and across regionsin 2000. The standard error of the gini index is estimated by jackknife andbootstrap methods rather than the traditional delta method because of thecomplexity of the latter.National gini coefficient of household current income increased by 23%from 0.3257 in 1996 to 0.3934 in 2000, while changes in regional giniindices ranged from -1.7% for Jeju to +35.6% for Geonggi-do. But, thechanges in gini index are not statistically significant at 10% significancelevel in three regions: Jeju(-1.7% from 0.3730 to 0.3667), Busan(0.74% from0.3799 to 0.3827) and Jeonnam( 4.99% from 0.3949 to 0.4146).Regional gini indices of household total income in 2000 ranged from0.3442 in Ulsan to 0.4528 in Gyeongbuk, but those of similar magnitude arenot statistically significant. For example, Busan’s income inequality(0.3852)does not differ significantly from 10 regions including Incheon(0.3614) toDaejeon(0.4107).