Abstract: A collaborative operation of cross-regional agricultural machinery has been one of the most important ways to improve the utilization rate for a trade-off between small farms and modern mechanized agriculture. In this study, a spatial flow mechanism was proposed for the cross-regional operation of wheat combined harvesters from the multi-scale perspective of "province-city-county" in China. The operational data of 16938 Lovol wheat combined harvesters were also collected from May to June 2020. Taking the specific provinces, cities, districts, and counties as network nodes, the flow of wheat combined harvesters was calculated for each node. After that, an ArcGIS software was used to map the inflow and outflow pattern, spatial and temporal distribution, inflow and outflow of the first degree at different nodes. A social network analysis was then utilized to determine the network structures and internal characteristics of the cross-regional operation of wheat combines, in terms of network density, centrality, and group clustering. The results showed that: 1) There was an outstanding heterogeneity of spatial-temporal flow under different scales. The flow volume of wheat combines was distributed in the form of a block, band, and core aggregation in turn, with the reduction of spatial scale. Specifically, there was a spatial connection pattern of "Henan-Hebei-Shandong" triangle at the provincial level. There was a scattering pattern at the municipal and prefectural level, where Nanyang, Zhumadian, Xianyang, and Xi'an were some of the main divergence cores. At the same time, there was a gradually increasing connection of the nodes at short and medium distances. 2) During the busy farming season of harvest and sowing in summer, the wheat combines remained moving with the maturity of wheat in different areas, where the concentrated center of cross-regional operation shifted roughly from the south to the north. Furthermore, there was an "M" shape pattern for the sum scales of cross-regional operation, indicating several peaks from mid-late May to mid-early June. 3) Henan province was the main departure and terminal of wheat combines. Moreover, the spatial orientation was relatively stronger as a terminal. 4) Henan, Hebei, and Shandong provinces were in the network center of cross-regional operation. There was also a high centrality in some prefecture-level cities, like Nanyang, Zhumadian, Zhoukou, Shangqiu, Baoding, Liaocheng, Heze, and Handan. 5) Among several sub-networks, there were the largest intra-operation scales in Henan and Hebei provinces, where the average distance was 248 and 238 km, respectively. There were also the most closely agglomerative sub-groups around the Henan and Hebei provinces. Consequently, this finding can greatly contribute to grasping the spatiotemporal flow for the cross-regional operation of agricultural machinery at the macroscopic level. As such, higher efficiency of cross-regional dispatch can be achieved for the decision-making on the whole field management. Furthermore, the cross-regional operation of wheat combines was characterized by the long overall duration, wide range, many participants, concentrated operation, and large flow in a certain area in a short time. Some recommendations were also made: 1) To optimize the cross-regional operation in the whole agriculture community; 2) To train the professional skills for the optimal operation route and quality with the aid of the modern machines and tools; 3) To improve after-sales service for the higher performance and reliability of agricultural machinery, particularly for the maintenance points in the large centralized operation areas, according to the flow of wheat combine harvesters; 4) To strengthen the information services of cross-regional operation, including the timely release of operational information for the orderly competition of agriculture markets.