Crop sensor-based in-season N management strategies have been successfully developed and evaluated for winter wheat around the world, but little has been reported for rice. The objective of this study was to develop an active crop sensor-based in-season N management strategy for upland rice in Northeast China.  Four field experiments were conducted in 2008 and 2009 in Jiansanjiang, Heilongjiang Province, Northeast China. The experiments used randomized block design with three or four replications and five N rate treatments: 0, 35, 70, 105 and 140 kg N/ha. The N fertilizers were split into three applications: 45% before transplanting, 20% at tillering stage, and 35% at panicle elongation. In order to estimate rice yield potential without N application at panicle elongation, each plot was split into two parts: the main plot and the split plot without receiving the third N application. An active crop canopy sensor, GreenSeeker, was used to collect rice canopy reflectance data five times across the growing season. Aboveground biomass, N concentration and uptake were determined following each sensor data collection. At harvest, rice grain yield, biomass, N concentration and uptake were also determined. GreenSeeker and yield data were also collected from a few farmer’s fields to evaluate the results from the N rate experiments. Preliminary analysis indicated that In-Season Estimated Yield (INSEY) was highly correlated with measured yield in plots without N application at panicle elongation at site 1, but the relationship was quite weak at panicle elongation stage at site 2.  Response index (RI) calculated with normalized difference vegetation index (NDVI) at panicle elongation was significantly correlated with RI calculated with yield at site 1. A preliminary N recommendation algorithm was developed and the potential application and future research needs will be discussed.