Forested watersheds provide fresh water with best quality and forestry practices play important role on the water production in the watersheds. Therefore, the objective of this study was to investigate the effect of 18 % thinning, which was under the 20 % threshold value reported in the literature on water yield in a forested watershed. Two experimental watersheds with similar ecological conditions in the Belgrad Forest of Istanbul were studied. Following a 6-year calibration period from December 2005 to October 2011, a simple linear regression equation was developed between the monthly streamflows of two watersheds with a significantly high correlation coefficient (r = 0.95). After 18 % of standing volume was removed from one watershed and the other was left untreated as a control, the streamflow was monitored for 2 years starting in January 2012 in both watersheds. The change in the monthly runoff was estimated as the difference between measured and values calculated with the linear regression equation. Average monthly streamflows were about 19, 15 and 10 mm in the control and 21, 20, and 11 mm in the treatment watersheds for calibration, first and second post-treatment periods, respectively. Paired watershed analysis showed that monthly streamflow did not significantly increase in either watershed for the first or the second year after the harvest. The results revealed that thinning intensity had to be greater to increase water yield significantly in this forest ecosystem and the threshold value for a streamflow increase was greater than 18 % for this region.