This thesis is using a non-parametric methodology, an extention of Linear Programming of what so called Data Envelopment Analysis (DEA). We estimate efficiency scores for DKI Jakarta public secondary high-shcool, using data mainly for year 2004 and 2005. The input measures are constructed from the new entrance student's minimum national examination score (NEM), annual school revenue per student from the government subsidies as well as from the public, while the outputs measures are based on the national examination (UAN) score of the respective students, UAN pass rate, and the rate of students entering public universities. Using DEA we are able to separate schools that might qualify, as "performing well" from those were some improvement might be possible in terms of efficiency, and to identify the benchmarks as well as to provide the input target needed by the inefficient school in order to reallocate their resource more efficiently. Furthermore we advance the use of non-parametric rank-sum test to prove whether the population that grouped by the intensity of school revenue from public per students, would positively correlated with the level of school performance in efficient term.