Despite their potential benefits, streamflow reconstructions from tree-rings have not been widely used in water systems analysis because the flow is generally constructed at annual resolution, which may be too coarse for analysis of drought/flood vulnerability and decision making. In semi-arid regions of Chile there are very good relationships between these daily and annual flow, which is explained because within these regions, few hydrological events within a year explain most of the annual flow. The understanding of the long-term evolution of peak flows has direct effects on public works planning, since the design of all infrastructure associated with water (including dams, bridges, etc.) rely on a risk assessment calculated from flood return periods and the work’s service life. Flood return periods are calculated from the streamflow maxima annual time series (i.e., one value per year). Unfortunately, instrumental records are often too short to make accurate estimation of flood recurrence intervals. In this context, we did the first streamflow reconstructions of maximum daily flow of the Limarí river in the semiarid climate, and the Petorca river in the Mediterranean region of Chile. The reconstructions explain the 59% and 52% of the total instrumental record variance, in both cases with good calibration and error statistics. These reconstructions are too short today, only 100 years, extending around 60 years the the instrumental records, but are not enough to compare the extreme return interval with the hydrological estimation. New tree-ring chronologies are been developing to solve this issue. We hope in the near future compare the new long-term records with the hydrological estimation to build infraestructure.