Rates of primary production were measured in 2 estuaries (Randers Fjord, Denmark, and the Scheldt estuary, Belgium/The Netherlands) using 3 different incubation methods: (1) the oxygen light-dark method (O2-LD), (2) 14C incorporation and (3) 18O labeling. Estimates based on the 14C incorporation technique were not significantly different from those obtained using the O2-LD technique. The 18O approach provided rates significantly lower than the 2 other techniques. Ratios of O2-LD to 18O-based rates (range: 0.99 to 3.54) were often statistically significantly higher than 1 and increased with decreasing salinities and/or lower oxygen concentrations. The underestimation of gross primary production by the 18O method may be due to an intracellular recycling of labeled oxygen which increased in magnitude with decreasing external oxygen conditions. These results suggest that the 18O method must be used with extreme care in nutrient-rich, low oxygen systems.