This study aimed to develop an instrument for assessing science learning outcomes based on HOTS (Higher Order Thinking Skills). This research was development research that uses a 4D model (four-D model) with development stages consisting of define, design, develop, and distribute. This development research was only carried out until the development stage. The subject of this research was the HOTS-based science learning outcome assessment instrument in the form of grids and multiple-choice test sheets. Data were obtained by using interviews, observation, and test methods. The validation of the assessment instrument was carried out by two material experts using a validation sheet and 78 students for a limited trial using multiple-choice objective test instruments. The results obtained will be analyzed for validity, reliability, discriminatory power, level of difficulty, and quality of distractors. The results of the analysis of the HOTS-based science learning outcome assessment instrument had a validity of 0.90 which was in the very high category, reliability of 0.81 which was in the very high category. Distinguishing analysis obtained 2 items with very good criteria, 14 items with good criteria, and 9 items with sufficient criteria. In the difficulty level test, 12 questions were obtained in the easy category and 13 questions in the medium category. Analysis of the distractor quality test found 63 distractors were at the >5% level, which means the distractors are functioning well and 12 distractors were at the 5% level, which means the distractors were not functioning properly. These results indicated that the HOTS-based science learning outcome test assessment instrument developed was valid and reliable and is suitable for use as an assessment instrument in various styles of material.