<![CDATA[This research purpose was to know the effect of discovery learning toward students critical thinking skill of SMPN 1 Gadingrejo. The design was pretest-protest non-equivalent. The sample were students of class VII2 and VII3, that were selected by purposive sampling. Data were obtained from pretest, posttest, and N-Gain, which were analyzed by t-test and u-test. Data of students activity which were obtained from observation sheet was analyzed descriptively. The result showed that N-Gain score of experimental class was 64,80 significantly different from control class which was 43,92. The average of critical thinking skill has criteria for each indicators. Giving basic explanation was very good, making basic skill was good, giving further explanation was good, and concluding was good. Therefore, it can be concluded that discovery learning was effected in improving students critical thinking skill. Tujuan penelitian ini untuk mengetahui pengaruh model discovery learning terhadap kemampuan berpikir kritis siswa SMPN 1 Gadingrejo. Desain penelitian adalah pretest-posttest kelompok tak ekuivalen. Sampel penelitian ini adalah siswa kelas VII2 dan VII3 yang dipilih secara purposive sampling. Data kemampuan berpikir kritis siswa yang diperoleh dari rata-rata nilai pretest, posttest dan N-gain, dianalisis menggunakan uji-t dan uji-u. Data aktivitas dari lembar observasi dan dianalisis secara deskriptif. Hasil penelitian menunjukkan kemampuan berpikir kritis siswa dari rata-rata N-gain kelas eksperimen sebesar 64,80 berbeda signifikan dengan kelas kontrol 43,92. Rata-rata aktivitas kemampuan berpikir kritis memiliki kriteria untuk setiap indikator. Memberikan penjelasan dasar berkriteria Sangat Baik, membangun keterampilan dasar berkriteria Baik, memberikan penjelasan lanjut berkriteria Baik, dan menyimpulkan berkriteria Baik. Dengan demikian, pembelajaran menggunakan model discovery learning berpengaruh dalam meningkatkan kemampuan berpikir kritis. Kata kunci : discovery learning, kemampuan berpikir kritis, pengaruh model]]>
