Abstract: Multiple goals are often required to balance the agricultural productivity for the food supply and the minimum resource input into the environmental damage in the ecological system services. Agricultural eco-efficiency has been used to evaluate the economic and ecological benefits for the better production performance. In this study, an assessment system was established to explore the spatial and temporal development of agricultural eco-efficiency. The panel data was collected from 678 towns in northern Jiangsu Province of China from 2000 to 2020. A data envelopment model was also used to assess the agricultural eco-efficiency using relaxation value measurement. In addition, the spatiotemporal evolution and regional differences of agricultural eco-efficiency were systematically investigated using global- and local-spatial autocorrelation analysis. Finally, the geographical detector model was selected to identify the influence of the component indicators on the spatial pattern of agricultural eco-efficiency. The results showed that: 1) The agricultural eco-efficiency shared a wave-like trend of "rise, fall, and rise". There was the great variation in the agricultural eco-efficiency in each urban area over the time. The descending order was ranked in Huai'an City>Yancheng City>northern Jiangsu Province>Lianyungang City>Suqian City>Xuzhou City. Among them, the maximum increase of agricultural eco-efficiency was found in Yancheng City, with an increase of 22.77%. By contrast, Suqian City decreased from 0.46 in 2000 to 0.35 in 2020, whereas, Huai'an City was basically the same from 2000 to 2020. The number of townships was 78, 98, 67, 57, and 88, respectively, with the high agricultural eco-efficiency (>0.81-1.0) in the study area. The townships with the high agricultural eco-efficiency were accounted for only 12.98% of the total in 2020, indicating the great potential to the improvement. 2) The overall agricultural eco-efficiency showed a spatial pattern of "high in the south and low in the north". The high-quality areas were small and scattered with a tendency to expand along the south-west to the north-east. By contrast, the low-quality areas were concentrated in Xuzhou, Suqian, and Lianyungang City. A clear pattern of agglomeration was observed with the low-low clustering region (23.89%-25.22%) in Xuzhou City, Ganyu County in the northern area of Lianyungang City, and Shuyang County in the northern of Suqian City, indicating the high spatial aggregation. The high-high clustering region (14.9%-24.34%) was shifted from the areas, such as Xuyi County in the north and south-west of Huai'an City to Yancheng City. 3) Energy inputs, pesticide inputs, and agricultural carbon emissions were the dominant factors in the spatial differentiation of agricultural eco-efficiency. The second most important factors were determined as the fertilizer inputs, agricultural film inputs, agricultural non-point source pollution, machinery inputs, and labor inputs in the spatial pattern of agricultural eco-efficiency. There was the different increase in the q-values for the interactions of each indicator. Two factors were also combined to strengthen the influence on the agricultural eco-efficiency. The type of interaction was also the non-linear strengthening (73%). Therefore, the agricultural carbon emissions, energy and machinery inputs were the common dominant factors using the factor detector. The weaker influence was found from the grain yield and ecosystem service function using factor detector. Anyway, the synergistic effects of various factors can be expected to achieved the optimal output on the spatial pattern of agricultural eco-efficiency. The finding can also provide a better guidance for the decision-making on the agricultural eco-efficiency in the environmentally friendly development of regional agriculture.