Stream runoff data globally and in the state of Minnesota were used to estimate subsurface water flow. This system approach is based, in principal, on unity of groundwater and surface water systems, and it is in stark contrast to the traditional deterministic approach based on modeling. In coordination with methodology of system analysis, two levels of study were used to estimate subsurface flow. First, the global stream runoff data were assessed to estimate the temporal-spatial variability of surface water runoff. Factor analysis was used to study the temporal-spatial variability of global runoff for the period from 1918 to 1967. Results of these analysis demonstrate that the variability of global runoff could be represented by seven major components (factor scores) that could be grouped into seven distinct independent grouping from the total of 18 continental slopes on the Earth. Computed variance value in this analysis is 76% and supports such analysis. The global stream runoff for this period is stationary, and is more closely connected with the stream flow of Asia to the Pacific Ocean as well as with the stream runoff of North America towards the Arctic and Pacific Oceans. The second level examines the distribution of river runoff (annual and for February) for various landscapes and the hydrogeological conditions in the State of Minnesota (218,000 km2). The annual and minimal monthly rate of stream runoff for 115 gauging stations with a period of observation of 47 years (1935-1981) were used to characterize the spatio-temporal distribution of stream runoff in Minnesota. Results of this analysis demonstrate that the annual stream runoff rate changes from 6.3, towards 3.95, and then to 2.09 1 s-1 km-2 (the difference is significant based on Student's criteria). These values in Minnesota correspond to ecological provinces from a mixed forest province towards the broadleaf forest and to prairie province, respectively. The distribution of minimal monthly stream runoff rate (February runoff) is controlled by hydrogeological systems in Minnesota. The difference between the two hydrogeological regions, Precambrian crystalline basement and Paleozoic artesian basin of 0.83 and 2.09 1/s/km2, is statistically significant. Within these regions, the monthly minimal runoff (0.5 and 1.68, and 0.87 and 3.11 1 s-1 km-2 for February, respectively) is also distinctly different for delineated subregions, depending on whether or not the Quaternary cover is present. The spatio-temporal structure that emerges could thus be used to generate river runoff and subsurface flow maps at any scale - from the global level to local detail. Such analysis was carried out in Minnesota with the detailed mapping of the subsurface flow for the Twin Cities Metropolitan area.