A method for real-time mosaic of video flow acquired by a small low-cost unmanned aerial vehicle (UAV) has been presented in this paper. The basic procedures of real-time mosaic are as follows: (1) Each video frame is resampled and orthorectified using a developed mathematical model, which can simultaneously solve the video camera's interior orientation parameters and the exterior orientation parameters of each video frame; (2) each orthorectified video frame is mosaicked at real time. A test field located in Picayune, Mississippi, has been established for testing our method. Sixty-minute video data were collected using the UAV and were processed using the proposed method. The results demonstrated that each video frame can be geo-orthorectified and mosaicked together to produce a 2-D planimetric mapping at near real time. Accuracy of the mosaicked video images (2-D planimetric map) is approximately 1-2 pixels, when compared to 55 checkpoints, which were measured by differential GPS surveying.