In the processing of infrared optical microstructured surfaces， the hard， brittle， and difficult-to-machine properties of infrared optical materials and the complex geometric properties of microstructured surfaces lead to uneven and brittle fractures in the processed microstructured surfaces and reduce the face shape accuracy. Small feeds are now commonly used to suppress surface fragmentation but are inefficient. To achieve efficient， high-precision， and low-damage machining of infrared optical microstructured surfaces， an ultra-precise adaptive flying cutting method was proposed and experimentally validated in this study. First， based on the kinematic characteristics of flying cutting， a flying cutting plasticity machining model was established. Second， based on the principle whereby the maximum chip thickness was always less than the brittle-plastic transition threshold， an iterative algorithm was used to plan a tool trajectory with dynamically varying feed rates based on the local morphological characteristics of the microstructured surface. Finally， the effectiveness of the proposed adaptive flying cutting method was verified by comparing it with the conventional flying cutting method in experiments. Experiments show that microgrooves are successfully machined without brittle fracture on single-crystal silicon materials and that a surface roughness of 18 nm is achieved. Compared with conventional flying tool cutting methods， the proposed ultra-precision adaptive flying cutting method avoids brittle breakage without reducing feed rates and achieves 2.5 times the machining efficiency of conventional methods.