Deforestation rates in Brazilian Amazon have been reaching around 20.000 km2.year-1. Around 80% of these deforested areas are converted to pasture. For this reason, within the Amazon region, pastures have been considered the main land cover category in hydrological studies with land cover changes. In order to assess the hydrological impact of conversion of forest to pasture, a comparative hydrological experiment was set up in the Central Amazon, close to Manaus city. Two small catchments of around 1.2 km2 size were instrumented, in a pasture and in a forest sites. Water level, precipitation and meteorological variables were continuously monitored during the study period. The frequency of measures was in time step of 30 minutes. Water level measurements were converted to discharge using a relationship built by salt dilution method and the evapotranspiration was estimated by Penman-Monteith equation for both catchments. Salt dilution method was satisfactory to measure discharge and to build a mathematical model to convert water level to discharge estimation. For the forest, an exponential model provided the best fit (R2 = 0.984; P<0.001), while for the pasture, a polynomial model was more adequate (R2 = 0.985; P < 0.001). During the study period, total rainfall was very similar in both catchments, with a difference of just 20 mm. This low difference validates all ensuing analysis. In spite of this, significant differences were found on the other water balance components. Evapotranspiration was 1.1 mm.day-1 lesser in the pasture catchment than in the forest. 50% of this difference (0.54 mm.day-1) is because of the canopy interception. Streamflow and water storage were respectively 0.98 mm.day-1 e 0.25 mm.day-1 higher in the pasture catchment than in the forest. The sum of these values is quite similar to evapotranspiration differences. In this sense, it is possible that the evapotranspiration is the main driver component of hydrological processes changes in tropical areas. Stormflow yield was 15.28% within forest site and 26.41% in the pasture site. Runoff coefficients (the relation of discharge over precipitation) were 0.17 and 0.32 for forest and pasture respectively. Overland flow was observed in the pasture site during intense rainfall events. Forest exerts a regulative effect on hydrological cycle, presenting uniform discharges, even over irregular precipitations. Analysis of individual events have shown significant differences between forest and pasture in terms of the rainfall/runoff response. The speed response to precipitation in the pasture site is higher than in the forest site. In spite of the similarity of precipitation levels, the discharge peaks are quicker and higher in pasture catchment. On the other hand, in the forest catchment, the recession is slower and storm flow yield is attenuated due to canopy interception and surface roughness. The flow duration curves showed more irregular behavior in the pasture catchment. Although discharge was higher in the pasture site during almost the whole period analyzed, discharge range is higher as well. Flow duration curves also pointed out that on the peak of dry season discharge was lower in pasture catchment. Forest lose a great part of available water by evapotranpiration, and exerts an important ecological role in the water balance cycles. Observations suggest that conversion of forest to pasture might have significant impacts on the ability to regularize floods during the wet season and droughts in dry seasons on larger scales.